Jokimbe-ComfyUI-DrawThings-gRPC/src-web/models/models.json

[
  {
    "version": "v1",
    "prefix": "redshift style ",
    "upcast_attention": false,
    "default_scale": 8,
    "file": "redshift_v1_f16.ckpt",
    "text_encoder": "redshift_v1_clip_vit_l14_f16.ckpt",
    "name": "3D Model (Redshift v1)"
  },
  {
    "name": "3D Model 768 (Redshift 768)",
    "objective": {
      "v": {}
    },
    "file": "redshift_768_v_f16.ckpt",
    "version": "v2",
    "upcast_attention": false,
    "prefix": "redshift style ",
    "text_encoder": "redshift_768_v_open_clip_vit_h14_f16.ckpt",
    "default_scale": 12
  },
  {
    "upcast_attention": false,
    "version": "v1",
    "name": "AloeVera's SimpMaker 3K1",
    "file": "aloeveras_simpmaker_3k1_f16.ckpt",
    "prefix": "",
    "default_scale": 8
  },
  {
    "name": "Analog (v1)",
    "prefix": "analog style ",
    "default_scale": 8,
    "version": "v1",
    "upcast_attention": false,
    "file": "analog_v1_f16.ckpt",
    "text_encoder": "analog_v1_clip_vit_l14_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Animagine XL v3.1",
    "modifier": "none",
    "text_encoder": "animagine_xl_v3.1_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "animagine_xl_v3.1_clip_vit_l14_f16.ckpt",
    "file": "animagine_xl_v3.1_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Animagine XL v3.1 (8-bit)",
    "modifier": "none",
    "text_encoder": "animagine_xl_v3.1_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "animagine_xl_v3.1_clip_vit_l14_f16.ckpt",
    "file": "animagine_xl_v3.1_q6p_q8p.ckpt"
  },
  {
    "prefix": "",
    "upcast_attention": false,
    "clip_encoder": "open_clip_vit_h14_visual_proj_f16.ckpt",
    "text_encoder": "open_clip_vit_h14_vision_model_f16.ckpt",
    "conditioning": "noise",
    "version": "svd_i2v",
    "default_scale": 8,
    "name": "AnimateLCM SVD XT v1.1",
    "noise_discretization": {
      "edm": {
        "_0": {
          "sigma_max": 700,
          "sigma_data": 0.5,
          "sigma_min": 0.002
        }
      }
    },
    "modifier": "none",
    "objective": {
      "v": {}
    },
    "file": "animatelcm_svd_xt_v1.1_f16.ckpt"
  },
  {
    "prefix": "",
    "upcast_attention": false,
    "clip_encoder": "open_clip_vit_h14_visual_proj_f16.ckpt",
    "text_encoder": "open_clip_vit_h14_vision_model_f16.ckpt",
    "conditioning": "noise",
    "version": "svd_i2v",
    "default_scale": 8,
    "name": "AnimateLCM SVD XT v1.1 (8-bit)",
    "noise_discretization": {
      "edm": {
        "_0": {
          "sigma_max": 700,
          "sigma_data": 0.5,
          "sigma_min": 0.002
        }
      }
    },
    "modifier": "none",
    "objective": {
      "v": {}
    },
    "file": "animatelcm_svd_xt_v1.1_q6p_q8p.ckpt"
  },
  {
    "text_encoder": "anything_v3_clip_vit_l14_f16.ckpt",
    "default_scale": 8,
    "version": "v1",
    "file": "anything_v3_f16.ckpt",
    "prefix": "",
    "autoencoder": "anything_v3_vae_f16.ckpt",
    "name": "Anime (Anything v3)",
    "upcast_attention": false
  },
  {
    "upcast_attention": false,
    "version": "v1",
    "file": "wd_v1.3_f16.ckpt",
    "prefix": "",
    "name": "Anime (Waifu Diffusion v1.3)",
    "default_scale": 8
  },
  {
    "name": "AniSora v3.2 I2V Wan 2.2 A14B High Noise Expert",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "anisora_v3.2_i2v_wan_2.2_a14b_hne_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "note": "[AniSora v3.2](https://huggingface.co/IndexTeam/Index-anisora) is a state-of-the-art image-to-video model fine-tuned by Bilibili for anime generation."
  },
  {
    "name": "AniSora v3.2 I2V Wan 2.2 A14B High Noise Expert (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "anisora_v3.2_i2v_wan_2.2_a14b_hne_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "note": "[AniSora v3.2](https://huggingface.co/IndexTeam/Index-anisora) is a state-of-the-art image-to-video model fine-tuned by Bilibili for anime generation.",
    "builtin_lora": true
  },
  {
    "name": "AniSora v3.2 I2V Wan 2.2 A14B Low Noise Expert",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "anisora_v3.2_i2v_wan_2.2_a14b_lne_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "note": "[AniSora v3.2](https://huggingface.co/IndexTeam/Index-anisora) is a state-of-the-art image-to-video model fine-tuned by Bilibili for anime generation."
  },
  {
    "name": "AniSora v3.2 I2V Wan 2.2 A14B Low Noise Expert (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "anisora_v3.2_i2v_wan_2.2_a14b_lne_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "note": "[AniSora v3.2](https://huggingface.co/IndexTeam/Index-anisora) is a state-of-the-art image-to-video model fine-tuned by Bilibili for anime generation.",
    "builtin_lora": true
  },
  {
    "file": "arcane_v3_f16.ckpt",
    "text_encoder": "arcane_v3_clip_vit_l14_f16.ckpt",
    "name": "Arcane (v3)",
    "default_scale": 8,
    "prefix": "arcane style ",
    "upcast_attention": false,
    "version": "v1"
  },
  {
    "name": "Artsy Vibe v1",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "artsy_vibe_v1_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "Artsy Vibe v1 (5-bit, SVDQuant)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "artsy_vibe_v1_q5p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "builtin_lora": true
  },
  {
    "name": "AuraFlow v0.1",
    "file": "auraflow_v0.1_q8p.ckpt",
    "prefix": "",
    "version": "auraflow",
    "defaultScale": 16,
    "textEncoder": "pile_t5_xl_encoder_q8p.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "deprecated": true
  },
  {
    "name": "AuraFlow v0.1 (8-bit)",
    "file": "auraflow_v0.1_q5p.ckpt",
    "prefix": "",
    "version": "auraflow",
    "defaultScale": 16,
    "textEncoder": "pile_t5_xl_encoder_q8p.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "deprecated": true
  },
  {
    "name": "AuraFlow v0.2",
    "file": "auraflow_v0.2_q8p.ckpt",
    "prefix": "",
    "version": "auraflow",
    "defaultScale": 16,
    "textEncoder": "pile_t5_xl_encoder_q8p.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt"
  },
  {
    "name": "AuraFlow v0.2 (8-bit)",
    "file": "auraflow_v0.2_q5p.ckpt",
    "prefix": "",
    "version": "auraflow",
    "defaultScale": 16,
    "textEncoder": "pile_t5_xl_encoder_q8p.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt"
  },
  {
    "name": "Balloon Art (v1)",
    "upcast_attention": false,
    "version": "v1",
    "file": "balloonart_v1_f16.ckpt",
    "prefix": "balloonart ",
    "default_scale": 8,
    "text_encoder": "balloonart_v1_clip_vit_l14_f16.ckpt"
  },
  {
    "name": "Chroma v47 Detail Calibrated",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "chroma_v47_detail_calibrated_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "mmdit": {
      "dual_attention_layers": [],
      "distilled_guidance_layers": 5,
      "qk_norm": true
    },
    "note": "See more about [Chroma](https://huggingface.co/lodestones/Chroma)."
  },
  {
    "name": "Chroma v47 Detail Calibrated (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "chroma_v47_detail_calibrated_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "mmdit": {
      "dual_attention_layers": [],
      "distilled_guidance_layers": 5,
      "qk_norm": true
    },
    "note": "See more about [Chroma](https://huggingface.co/lodestones/Chroma)."
  },
  {
    "name": "Chroma v48 Detail Calibrated",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "chroma_v48_detail_calibrated_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "mmdit": {
      "dual_attention_layers": [],
      "distilled_guidance_layers": 5,
      "qk_norm": true
    },
    "note": "See more about [Chroma](https://huggingface.co/lodestones/Chroma)."
  },
  {
    "name": "Chroma v48 Detail Calibrated (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "chroma_v48_detail_calibrated_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "mmdit": {
      "dual_attention_layers": [],
      "distilled_guidance_layers": 5,
      "qk_norm": true
    },
    "note": "See more about [Chroma](https://huggingface.co/lodestones/Chroma)."
  },
  {
    "name": "Chroma1 HD",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "chroma_1_hd_r0.1_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "mmdit": {
      "dual_attention_layers": [],
      "distilled_guidance_layers": 5,
      "qk_norm": true
    },
    "note": "See more about [Chroma1 HD](https://huggingface.co/lodestones/Chroma1-HD)."
  },
  {
    "name": "Chroma1 HD (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "chroma_1_hd_r0.1_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "mmdit": {
      "dual_attention_layers": [],
      "distilled_guidance_layers": 5,
      "qk_norm": true
    },
    "note": "See more about [Chroma1 HD](https://huggingface.co/lodestones/Chroma1-HD)."
  },
  {
    "name": "ChronoEdit 14B",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 15,
    "hires_fix_scale": 16,
    "file": "chronoedit_14b_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "note": "[ChronoEdit 14B](https://research.nvidia.com/labs/toronto-ai/chronoedit/) is a image edit model fine-tuned from Wan 2.1 14B video model, that prefers 29-frame or 5-frame generation."
  },
  {
    "name": "ChronoEdit 14B (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 15,
    "hires_fix_scale": 16,
    "file": "chronoedit_14b_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "note": "[ChronoEdit 14B](https://research.nvidia.com/labs/toronto-ai/chronoedit/) is a image edit model fine-tuned from Wan 2.1 14B video model, that prefers 29-frame or 5-frame generation.",
    "builtin_lora": true
  },
  {
    "text_encoder": "classicanim_v1_clip_vit_l14_f16.ckpt",
    "default_scale": 8,
    "version": "v1",
    "file": "classicanim_v1_f16.ckpt",
    "prefix": "classic disney style ",
    "name": "Classic Animation (v1)",
    "upcast_attention": false
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "ColorfulXL v6.0",
    "modifier": "none",
    "clip_encoder": "colorfulxl_v6.0_clip_vit_l14_f16.ckpt",
    "file": "colorfulxl_v6.0_f16.ckpt",
    "text_encoder": "colorfulxl_v6.0_open_clip_vit_bigg14_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "ColorfulXL v6.0 (8-bit)",
    "modifier": "none",
    "clip_encoder": "colorfulxl_v6.0_clip_vit_l14_f16.ckpt",
    "file": "colorfulxl_v6.0_q6p_q8p.ckpt",
    "text_encoder": "colorfulxl_v6.0_open_clip_vit_bigg14_f16.ckpt"
  },
  {
    "version": "v1",
    "default_scale": 8,
    "prefix": "",
    "name": "Counterfeit v3.0",
    "upcast_attention": false,
    "modifier": "none",
    "file": "counterfeit_v3.0_f16.ckpt",
    "text_encoder": "counterfeit_v3.0_clip_vit_l14_f16.ckpt"
  },
  {
    "version": "v1",
    "default_scale": 8,
    "prefix": "",
    "name": "Counterfeit v3.0 (8-bit)",
    "upcast_attention": false,
    "modifier": "none",
    "file": "counterfeit_v3.0_q6p_q8p.ckpt",
    "text_encoder": "counterfeit_v3.0_clip_vit_l14_f16.ckpt"
  },
  {
    "prefix": "dgs illustration style ",
    "file": "cyberpunk_anime_f16.ckpt",
    "default_scale": 8,
    "upcast_attention": false,
    "version": "v1",
    "name": "Cyberpunk Anime",
    "text_encoder": "cyberpunk_anime_clip_vit_l14_f16.ckpt"
  },
  {
    "version": "v1",
    "name": "Deliberate v2.0 (8-bit)",
    "file": "deliberate_v2_q6p_q8p.ckpt",
    "prefix": "",
    "text_encoder": "deliberate_v2_clip_vit_l14_f16.ckpt",
    "default_scale": 8,
    "upcast_attention": false
  },
  {
    "name": "devMODE v0.3 Turbo",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "devmode_v0.3_turbo_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "devMODE v0.3 Turbo (5-bit, SVDQuant)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "devmode_v0.3_turbo_q5p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "builtin_lora": true
  },
  {
    "default_scale": 8,
    "text_encoder": "disney_pixar_cartoon_type_b_clip_vit_l14_f16.ckpt",
    "upcast_attention": false,
    "prefix": "",
    "file": "disney_pixar_cartoon_type_b_q6p_q8p.ckpt",
    "name": "Disney Pixar Cartoon Type B (8-bit)",
    "version": "v1"
  },
  {
    "file": "dnd_classes_and_species_f16.ckpt",
    "prefix": "",
    "upcast_attention": false,
    "default_scale": 8,
    "version": "v1",
    "text_encoder": "dnd_classes_and_species_clip_vit_l14_f16.ckpt",
    "name": "DnD Classes and Species"
  },
  {
    "version": "v1",
    "default_scale": 12,
    "prefix": "",
    "name": "DreamShaper v8",
    "upcast_attention": false,
    "file": "dreamshaper_v8_f16.ckpt",
    "text_encoder": "dreamshaper_v8_clip_vit_l14_f16.ckpt",
    "modifier": "none"
  },
  {
    "version": "v1",
    "default_scale": 12,
    "prefix": "",
    "name": "DreamShaper v8 (8-bit)",
    "upcast_attention": false,
    "file": "dreamshaper_v8_q6p_q8p.ckpt",
    "text_encoder": "dreamshaper_v8_clip_vit_l14_f16.ckpt",
    "modifier": "none"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "DreamShaper XL v2.1 Turbo",
    "modifier": "none",
    "clip_encoder": "dreamshaper_xl_v2.1_turbo_clip_vit_l14_f16.ckpt",
    "file": "dreamshaper_xl_v2.1_turbo_f16.ckpt",
    "text_encoder": "dreamshaper_xl_v2.1_turbo_open_clip_vit_bigg14_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "DreamShaper XL v2.1 Turbo (8-bit)",
    "modifier": "none",
    "clip_encoder": "dreamshaper_xl_v2.1_turbo_clip_vit_l14_f16.ckpt",
    "file": "dreamshaper_xl_v2.1_turbo_q6p_q8p.ckpt",
    "text_encoder": "dreamshaper_xl_v2.1_turbo_open_clip_vit_bigg14_f16.ckpt"
  },
  {
    "version": "v1",
    "file": "dnd_30000_f16.ckpt",
    "prefix": "",
    "default_scale": 8,
    "upcast_attention": false,
    "name": "Dungeons and Diffusion (30000)",
    "text_encoder": "dnd_30000_clip_vit_l14_f16.ckpt"
  },
  {
    "prefix": "elden ring style ",
    "upcast_attention": false,
    "name": "Elden Ring (v3)",
    "file": "eldenring_v3_f16.ckpt",
    "default_scale": 8,
    "text_encoder": "eldenring_v3_clip_vit_l14_f16.ckpt",
    "version": "v1"
  },
  {
    "upcast_attention": false,
    "version": "v1",
    "name": "F222",
    "file": "f222_f16.ckpt",
    "prefix": "",
    "default_scale": 8
  },
  {
    "name": "FLUX.1 [dev]",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_dev_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "FLUX.1 [dev] (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_dev_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "FLUX.1 [dev] (Exact)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_dev_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "FLUX.1 [dev] De-distill",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_dev_de_distill_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "FLUX.1 [dev] De-distill (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_dev_de_distill_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "FLUX.1 [schnell]",
    "file": "flux_1_schnell_q8p.ckpt",
    "prefix": "",
    "version": "flux1",
    "defaultScale": 16,
    "textEncoder": "t5_xxl_encoder_q6p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "clipEncoder": "clip_vit_l14_f16.ckpt",
    "highPrecisionAutoencoder": true,
    "isConsistencyModel": true,
    "paddedTextEncodingLength": 256,
    "hiresFixScale": 24
  },
  {
    "name": "FLUX.1 [schnell] (5-bit)",
    "file": "flux_1_schnell_q5p.ckpt",
    "prefix": "",
    "version": "flux1",
    "defaultScale": 16,
    "textEncoder": "t5_xxl_encoder_q6p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "clipEncoder": "clip_vit_l14_f16.ckpt",
    "highPrecisionAutoencoder": true,
    "isConsistencyModel": true,
    "paddedTextEncodingLength": 256,
    "hiresFixScale": 24
  },
  {
    "name": "FLUX.1 Canny [dev]",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "modifier": "canny",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_canny_dev_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "FLUX.1 Canny [dev] (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "modifier": "canny",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_canny_dev_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "FLUX.1 Depth [dev]",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "modifier": "depth",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_depth_dev_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "FLUX.1 Depth [dev] (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "modifier": "depth",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_depth_dev_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "FLUX.1 Fill [dev]",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "modifier": "inpainting",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_fill_dev_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 128,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "FLUX.1 Fill [dev] (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "modifier": "inpainting",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_fill_dev_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 128,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "FLUX.1 Kontext [dev]",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "modifier": "kontext",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_kontext_dev_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "FLUX.1 Kontext [dev] (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "modifier": "kontext",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_kontext_dev_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "FLUX.1 Kontext [dev] (Exact)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "modifier": "kontext",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_kontext_dev_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "FLUX.1 Krea [dev]",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_krea_dev_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "FLUX.1 Krea [dev] (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "flux_1_krea_dev_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "Fooocus Inpaint SDXL v2.6",
    "file": "fooocus_inpaint_sd_xl_v2.6_f16.ckpt",
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "defaultScale": 16,
    "textEncoder": "open_clip_vit_bigg14_f16.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "modifier": "inpainting",
    "clipEncoder": "clip_vit_l14_f16.ckpt"
  },
  {
    "name": "Fooocus Inpaint SDXL v2.6 (8-bit)",
    "file": "fooocus_inpaint_sd_xl_v2.6_q6p_q8p.ckpt",
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "defaultScale": 16,
    "textEncoder": "open_clip_vit_bigg14_f16.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "modifier": "inpainting",
    "clipEncoder": "clip_vit_l14_f16.ckpt"
  },
  {
    "upcast_attention": false,
    "prefix": "ghibli style ",
    "text_encoder": "ghibli_v1_clip_vit_l14_f16.ckpt",
    "default_scale": 8,
    "file": "ghibli_v1_f16.ckpt",
    "name": "Ghibli (v1)",
    "version": "v1"
  },
  {
    "file": "hna_3dkx_1.1_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "hna_3dkx_1.1_clip_vit_l14_f16.ckpt",
    "default_scale": 8,
    "version": "v1",
    "name": "H&A's 3DKX 1.1",
    "prefix": "a 3d render / cartoon of "
  },
  {
    "version": "v1",
    "file": "hassanblend_v1.5.1.2_f16.ckpt",
    "prefix": "",
    "text_encoder": "hassanblend_v1.5.1.2_clip_vit_l14_f16.ckpt",
    "upcast_attention": false,
    "default_scale": 8,
    "name": "Hassanblend (v1.5.1.2)"
  },
  {
    "name": "HiDream E1 [full]",
    "file": "hidream_e1_full_q8p.ckpt",
    "prefix": "",
    "version": "hidream_i1",
    "defaultScale": 12,
    "textEncoder": "llama_3.1_8b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "modifier": "editing",
    "clipEncoder": "long_clip_vit_l14_f16.ckpt",
    "additionalClipEncoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "t5Encoder": "t5_xxl_encoder_q6p.ckpt",
    "highPrecisionAutoencoder": true,
    "paddedTextEncodingLength": 128,
    "hiresFixScale": 24,
    "note": "[HiDream-E1 [full]](https://huggingface.co/HiDream-ai/HiDream-E1-Full) is an image editing model built on HiDream-I1. It is MIT-licensed and commercially friendly. Trained at 768×768 resolution using a Flow Matching objective, the model performs best with trailing samplers and 30–50 sampling steps. For optimal results, ensure the width is set to 768 and use the following prompt format: Editing Instruction: {}. Target Image Description: {}."
  },
  {
    "name": "HiDream E1 [full] (5-bit)",
    "file": "hidream_e1_full_q5p.ckpt",
    "prefix": "",
    "version": "hidream_i1",
    "defaultScale": 12,
    "textEncoder": "llama_3.1_8b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "modifier": "editing",
    "clipEncoder": "long_clip_vit_l14_f16.ckpt",
    "additionalClipEncoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "t5Encoder": "t5_xxl_encoder_q6p.ckpt",
    "highPrecisionAutoencoder": true,
    "paddedTextEncodingLength": 128,
    "hiresFixScale": 24,
    "note": "[HiDream-E1 [full]](https://huggingface.co/HiDream-ai/HiDream-E1-Full) is an image editing model built on HiDream-I1. It is MIT-licensed and commercially friendly. Trained at 768×768 resolution using a Flow Matching objective, the model performs best with trailing samplers and 30–50 sampling steps. For optimal results, ensure the width is set to 768 and use the following prompt format: Editing Instruction: {}. Target Image Description: {}."
  },
  {
    "name": "HiDream E1-1",
    "file": "hidream_e1_1_q8p.ckpt",
    "prefix": "",
    "version": "hidream_i1",
    "defaultScale": 16,
    "textEncoder": "llama_3.1_8b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "modifier": "editing",
    "clipEncoder": "long_clip_vit_l14_f16.ckpt",
    "additionalClipEncoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "t5Encoder": "t5_xxl_encoder_q6p.ckpt",
    "highPrecisionAutoencoder": true,
    "paddedTextEncodingLength": 128,
    "hiresFixScale": 24,
    "note": "[HiDream-E1-1](https://huggingface.co/HiDream-ai/HiDream-E1-1) is an image editing model built on HiDream-I1. It is MIT-licensed and commercially friendly. Trained with dynamic resolutions (around 1MP) using a Flow Matching objective, the model performs best with trailing samplers and 30–50 sampling steps."
  },
  {
    "name": "HiDream E1-1 (5-bit)",
    "file": "hidream_e1_1_q5p.ckpt",
    "prefix": "",
    "version": "hidream_i1",
    "defaultScale": 16,
    "textEncoder": "llama_3.1_8b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "modifier": "editing",
    "clipEncoder": "long_clip_vit_l14_f16.ckpt",
    "additionalClipEncoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "t5Encoder": "t5_xxl_encoder_q6p.ckpt",
    "highPrecisionAutoencoder": true,
    "paddedTextEncodingLength": 128,
    "hiresFixScale": 24,
    "note": "[HiDream-E1-1](https://huggingface.co/HiDream-ai/HiDream-E1-1) is an image editing model built on HiDream-I1. It is MIT-licensed and commercially friendly. Trained with dynamic resolutions (around 1MP) using a Flow Matching objective, the model performs best with trailing samplers and 30–50 sampling steps."
  },
  {
    "name": "HiDream E1-1 (Exact)",
    "version": "hidream_i1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "modifier": "editing",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "hidream_e1_1_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "llama_3.1_8b_instruct_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_f16.ckpt",
    "clip_encoder": "long_clip_vit_l14_f16.ckpt",
    "additional_clip_encoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "high_precision_autoencoder": true,
    "padded_text_encoding_length": 128,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "note": "[HiDream-E1-1](https://huggingface.co/HiDream-ai/HiDream-E1-1) is an image editing model built on HiDream-I1. It is MIT-licensed and commercially friendly. Trained at dynamic resolutions (around 1MP) using a Flow Matching objective, the model performs best with trailing samplers and 30–50 sampling steps."
  },
  {
    "name": "HiDream I1 [dev]",
    "file": "hidream_i1_dev_q8p.ckpt",
    "prefix": "",
    "version": "hidream_i1",
    "defaultScale": 16,
    "textEncoder": "llama_3.1_8b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "clipEncoder": "long_clip_vit_l14_f16.ckpt",
    "additionalClipEncoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "t5Encoder": "t5_xxl_encoder_q6p.ckpt",
    "highPrecisionAutoencoder": true,
    "guidanceEmbed": true,
    "paddedTextEncodingLength": 128,
    "hiresFixScale": 24,
    "note": "[HiDream-I1 [dev]](https://huggingface.co/HiDream-ai/HiDream-I1-Dev) is a state-of-the-art open-source image generation model known for its strong prompt adherence across diverse styles, including photorealistic, cartoon, and artistic. It is MIT-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 20–30 sampling steps recommended. Text guidance is not effective for this model."
  },
  {
    "name": "HiDream I1 [dev] (5-bit)",
    "file": "hidream_i1_dev_q5p.ckpt",
    "prefix": "",
    "version": "hidream_i1",
    "defaultScale": 16,
    "textEncoder": "llama_3.1_8b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "clipEncoder": "long_clip_vit_l14_f16.ckpt",
    "additionalClipEncoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "t5Encoder": "t5_xxl_encoder_q6p.ckpt",
    "highPrecisionAutoencoder": true,
    "guidanceEmbed": true,
    "paddedTextEncodingLength": 128,
    "hiresFixScale": 24,
    "note": "[HiDream-I1 [dev]](https://huggingface.co/HiDream-ai/HiDream-I1-Dev) is a state-of-the-art open-source image generation model known for its strong prompt adherence across diverse styles, including photorealistic, cartoon, and artistic. It is MIT-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 20–30 sampling steps recommended. Text guidance is not effective for this model."
  },
  {
    "name": "HiDream I1 [dev] (Exact)",
    "version": "hidream_i1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "hidream_i1_dev_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "llama_3.1_8b_instruct_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_f16.ckpt",
    "clip_encoder": "long_clip_vit_l14_f16.ckpt",
    "additional_clip_encoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 128,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "note": "[HiDream-I1 [dev]](https://huggingface.co/HiDream-ai/HiDream-I1-Dev) is a state-of-the-art open-source image generation model known for its strong prompt adherence across diverse styles, including photorealistic, cartoon, and artistic. It is MIT-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 20–30 sampling steps recommended. Text guidance is not effective for this model."
  },
  {
    "name": "HiDream I1 [fast]",
    "file": "hidream_i1_fast_q8p.ckpt",
    "prefix": "",
    "version": "hidream_i1",
    "defaultScale": 16,
    "textEncoder": "llama_3.1_8b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "clipEncoder": "long_clip_vit_l14_f16.ckpt",
    "additionalClipEncoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "t5Encoder": "t5_xxl_encoder_q6p.ckpt",
    "highPrecisionAutoencoder": true,
    "isConsistencyModel": true,
    "paddedTextEncodingLength": 128,
    "hiresFixScale": 24,
    "note": "[HiDream-I1 [fast]](https://huggingface.co/HiDream-ai/HiDream-I1-Fast) is a state-of-the-art open-source image generation model known for its strong prompt adherence across diverse styles, including photorealistic, cartoon, and artistic. It is MIT-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 10–20 sampling steps recommended. Text guidance is not effective for this model."
  },
  {
    "name": "HiDream I1 [fast] (5-bit)",
    "file": "hidream_i1_fast_q5p.ckpt",
    "prefix": "",
    "version": "hidream_i1",
    "defaultScale": 16,
    "textEncoder": "llama_3.1_8b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "clipEncoder": "long_clip_vit_l14_f16.ckpt",
    "additionalClipEncoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "t5Encoder": "t5_xxl_encoder_q6p.ckpt",
    "highPrecisionAutoencoder": true,
    "isConsistencyModel": true,
    "paddedTextEncodingLength": 128,
    "hiresFixScale": 24,
    "note": "[HiDream-I1 [fast]](https://huggingface.co/HiDream-ai/HiDream-I1-Fast) is a state-of-the-art open-source image generation model known for its strong prompt adherence across diverse styles, including photorealistic, cartoon, and artistic. It is MIT-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 10–20 sampling steps recommended. Text guidance is not effective for this model."
  },
  {
    "name": "HiDream I1 [fast] (Exact)",
    "version": "hidream_i1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "hidream_i1_fast_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "llama_3.1_8b_instruct_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_f16.ckpt",
    "clip_encoder": "long_clip_vit_l14_f16.ckpt",
    "additional_clip_encoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 128,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "note": "[HiDream-I1 [fast]](https://huggingface.co/HiDream-ai/HiDream-I1-Fast) is a state-of-the-art open-source image generation model known for its strong prompt adherence across diverse styles, including photorealistic, cartoon, and artistic. It is MIT-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 10–20 sampling steps recommended. Text guidance is not effective for this model."
  },
  {
    "name": "HiDream I1 [full]",
    "file": "hidream_i1_full_q8p.ckpt",
    "prefix": "",
    "version": "hidream_i1",
    "defaultScale": 16,
    "textEncoder": "llama_3.1_8b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "clipEncoder": "long_clip_vit_l14_f16.ckpt",
    "additionalClipEncoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "t5Encoder": "t5_xxl_encoder_q6p.ckpt",
    "highPrecisionAutoencoder": true,
    "paddedTextEncodingLength": 128,
    "hiresFixScale": 24,
    "note": "[HiDream-I1 [full]](https://huggingface.co/HiDream-ai/HiDream-I1-Full) is a state-of-the-art open-source image generation model known for its exceptional prompt adherence across a wide range of styles, including photorealistic, cartoon, and artistic. It is MIT-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended."
  },
  {
    "name": "HiDream I1 [full] (5-bit)",
    "file": "hidream_i1_full_q5p.ckpt",
    "prefix": "",
    "version": "hidream_i1",
    "defaultScale": 16,
    "textEncoder": "llama_3.1_8b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "clipEncoder": "long_clip_vit_l14_f16.ckpt",
    "additionalClipEncoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "t5Encoder": "t5_xxl_encoder_q6p.ckpt",
    "highPrecisionAutoencoder": true,
    "paddedTextEncodingLength": 128,
    "hiresFixScale": 24,
    "note": "[HiDream-I1 [full]](https://huggingface.co/HiDream-ai/HiDream-I1-Full) is a state-of-the-art open-source image generation model known for its exceptional prompt adherence across a wide range of styles, including photorealistic, cartoon, and artistic. It is MIT-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended."
  },
  {
    "name": "HiDream I1 [full] (Exact)",
    "version": "hidream_i1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "hidream_i1_full_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "llama_3.1_8b_instruct_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_f16.ckpt",
    "clip_encoder": "long_clip_vit_l14_f16.ckpt",
    "additional_clip_encoders": [
      "long_open_clip_vit_bigg14_f16.ckpt"
    ],
    "high_precision_autoencoder": true,
    "padded_text_encoding_length": 128,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "note": "[HiDream-I1 [full]](https://huggingface.co/HiDream-ai/HiDream-I1-Full) is a state-of-the-art open-source image generation model known for its exceptional prompt adherence across a wide range of styles, including photorealistic, cartoon, and artistic. It is MIT-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended."
  },
  {
    "name": "Hunyuan Video T2V 720p",
    "version": "hunyuan_video",
    "autoencoder": "hunyuan_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 24,
    "file": "hunyuan_video_t2v_720p_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "llava_llama_3_8b_v1.1_q8p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": false,
    "guidance_embed": true,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 30,
    "tea_cache_coefficients": [
      733.226126,
      -401.131952,
      67.5869174,
      -3.149878,
      0.0961237896
    ],
    "note": "[Hunyuan Video](https://huggingface.co/tencent/HunyuanVideo) is a state-of-the-art text-to-video model developed by Tencent. It can generate video clips of up to 5 seconds in length. The recommended resolutions are 960×544 or 1280×720. The model supports up to 129 frames, with a recommended shift value of 7.0."
  },
  {
    "name": "Hunyuan Video T2V 720p (5-bit, SVDQuant)",
    "version": "hunyuan_video",
    "autoencoder": "hunyuan_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 24,
    "file": "hunyuan_video_t2v_720p_q5p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "llava_llama_3_8b_v1.1_q8p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": false,
    "guidance_embed": true,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 30,
    "tea_cache_coefficients": [
      733.226126,
      -401.131952,
      67.5869174,
      -3.149878,
      0.0961237896
    ],
    "note": "[Hunyuan Video](https://huggingface.co/tencent/HunyuanVideo) is a state-of-the-art text-to-video model developed by Tencent. It can generate video clips of up to 5 seconds in length. The recommended resolutions are 960×544 or 1280×720. The model supports up to 129 frames, with a recommended shift value of 7.0.",
    "builtin_lora": true
  },
  {
    "name": "iCatcher Cartoon",
    "file": "icatcher_cartoon_f16.ckpt",
    "prefix": "",
    "modifier": "none",
    "default_scale": 16,
    "upcast_attention": false,
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "clip_encoder": "icatcher_cartoon_clip_vit_l14_f16.ckpt",
    "version": "sdxl_base_v0.9",
    "text_encoder": "icatcher_cartoon_open_clip_vit_bigg14_f16.ckpt"
  },
  {
    "name": "iCatcher Cartoon (8-bit)",
    "file": "icatcher_cartoon_q6p_q8p.ckpt",
    "prefix": "",
    "modifier": "none",
    "default_scale": 16,
    "upcast_attention": false,
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "clip_encoder": "icatcher_cartoon_clip_vit_l14_f16.ckpt",
    "version": "sdxl_base_v0.9",
    "text_encoder": "icatcher_cartoon_open_clip_vit_bigg14_f16.ckpt"
  },
  {
    "name": "iCatcher Realistic",
    "file": "icatcher_realistic_f16.ckpt",
    "prefix": "",
    "modifier": "none",
    "default_scale": 16,
    "upcast_attention": false,
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "text_encoder": "icatcher_realistic_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "icatcher_realistic_clip_vit_l14_f16.ckpt",
    "version": "sdxl_base_v0.9"
  },
  {
    "name": "iCatcher Realistic (8-bit)",
    "file": "icatcher_realistic_q6p_q8p.ckpt",
    "prefix": "",
    "modifier": "none",
    "default_scale": 16,
    "upcast_attention": false,
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "text_encoder": "icatcher_realistic_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "icatcher_realistic_clip_vit_l14_f16.ckpt",
    "version": "sdxl_base_v0.9"
  },
  {
    "version": "v1",
    "file": "inkpunk_v2_f16.ckpt",
    "name": "Inkpunk (v2)",
    "upcast_attention": false,
    "prefix": "nvinkpunk ",
    "default_scale": 8,
    "text_encoder": "inkpunk_v2_clip_vit_l14_f16.ckpt"
  },
  {
    "name": "Instruct Pix2Pix",
    "file": "instruct_pix2pix_22000_f16.ckpt",
    "prefix": "",
    "version": "v1",
    "defaultScale": 8,
    "modifier": "editing",
    "deprecated": true
  },
  {
    "version": "v1",
    "default_scale": 12,
    "prefix": "",
    "name": "Juggernaut Reborn",
    "upcast_attention": false,
    "file": "juggernaut_reborn_f16.ckpt",
    "text_encoder": "juggernaut_reborn_clip_vit_l14_f16.ckpt",
    "modifier": "none"
  },
  {
    "version": "v1",
    "default_scale": 12,
    "prefix": "",
    "name": "Juggernaut Reborn (8-bit)",
    "upcast_attention": false,
    "file": "juggernaut_reborn_q6p_q8p.ckpt",
    "text_encoder": "juggernaut_reborn_clip_vit_l14_f16.ckpt",
    "modifier": "none"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Juggernaut XL Ragnarok",
    "modifier": "none",
    "text_encoder": "juggernaut_xl_ragnarok_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "juggernaut_xl_ragnarok_clip_vit_l14_f16.ckpt",
    "file": "juggernaut_xl_ragnarok_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Juggernaut XL Ragnarok (8-bit)",
    "modifier": "none",
    "text_encoder": "juggernaut_xl_ragnarok_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "juggernaut_xl_ragnarok_clip_vit_l14_f16.ckpt",
    "file": "juggernaut_xl_ragnarok_q6p_q8p.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Juggernaut XL v9",
    "modifier": "none",
    "text_encoder": "juggernaut_xl_v9_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "juggernaut_xl_v9_clip_vit_l14_f16.ckpt",
    "file": "juggernaut_xl_v9_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Juggernaut XL v9 (8-bit)",
    "modifier": "none",
    "text_encoder": "juggernaut_xl_v9_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "juggernaut_xl_v9_clip_vit_l14_f16.ckpt",
    "file": "juggernaut_xl_v9_q6p_q8p.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Juggernaut XL v9 Lightning",
    "modifier": "none",
    "file": "juggernaut_xl_v9_lightning_f16.ckpt",
    "text_encoder": "juggernaut_xl_v9_lightning_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "juggernaut_xl_v9_lightning_clip_vit_l14_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Juggernaut XL v9 Lightning (8-bit)",
    "modifier": "none",
    "file": "juggernaut_xl_v9_lightning_q6p_q8p.ckpt",
    "text_encoder": "juggernaut_xl_v9_lightning_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "juggernaut_xl_v9_lightning_clip_vit_l14_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Juggernaut XL X",
    "modifier": "none",
    "text_encoder": "juggernaut_xl_x_open_clip_vit_bigg14_f16.ckpt",
    "file": "juggernaut_xl_x_f16.ckpt",
    "clip_encoder": "juggernaut_xl_x_clip_vit_l14_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Juggernaut XL X (8-bit)",
    "modifier": "none",
    "text_encoder": "juggernaut_xl_x_open_clip_vit_bigg14_f16.ckpt",
    "file": "juggernaut_xl_x_q6p_q8p.ckpt",
    "clip_encoder": "juggernaut_xl_x_clip_vit_l14_f16.ckpt"
  },
  {
    "name": "Kandinsky v2.1",
    "file": "kandinsky_f16.ckpt",
    "prefix": "",
    "version": "kandinsky2.1",
    "upcastAttention": false,
    "defaultScale": 12,
    "textEncoder": "xlm_roberta_f16.ckpt",
    "autoencoder": "kandinsky_movq_f16.ckpt",
    "deprecated": true,
    "imageEncoder": "image_vit_l14_f16.ckpt",
    "clipEncoder": "clip_vit_l14_f16.ckpt",
    "diffusionMapping": "kandinsky_diffusion_mapping_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Kwai Kolors 1.0",
    "modifier": "none",
    "file": "kwai_kolors_1.0_f16.ckpt",
    "text_encoder": "chatglm3_6b_q6p_q8p.ckpt",
    "text_encoder_version": "chatglm3_6b"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Kwai Kolors 1.0 (8-bit)",
    "modifier": "none",
    "file": "kwai_kolors_1.0_q6p_q8p.ckpt",
    "text_encoder": "chatglm3_6b_q6p_q8p.ckpt",
    "text_encoder_version": "chatglm3_6b"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": true,
    "name": "Kwai Kolors Inpainting 1.0",
    "modifier": "inpainting",
    "file": "kwai_kolors_inpainting_1.0_f16.ckpt",
    "text_encoder": "chatglm3_6b_q6p_q8p.ckpt",
    "text_encoder_version": "chatglm3_6b"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": true,
    "name": "Kwai Kolors Inpainting 1.0 (8-bit)",
    "modifier": "inpainting",
    "file": "kwai_kolors_inpainting_1.0_q6p_q8p.ckpt",
    "text_encoder": "chatglm3_6b_q6p_q8p.ckpt",
    "text_encoder_version": "chatglm3_6b"
  },
  {
    "version": "v1",
    "default_scale": 8,
    "prefix": "",
    "name": "MajicMIX Realistic v7",
    "upcast_attention": false,
    "file": "majicmix_realistic_v7_f16.ckpt",
    "text_encoder": "majicmix_realistic_v7_clip_vit_l14_f16.ckpt",
    "modifier": "none"
  },
  {
    "version": "v1",
    "default_scale": 8,
    "prefix": "",
    "name": "MajicMIX Realistic v7 (8-bit)",
    "upcast_attention": false,
    "file": "majicmix_realistic_v7_q6p_q8p.ckpt",
    "text_encoder": "majicmix_realistic_v7_clip_vit_l14_f16.ckpt",
    "modifier": "none"
  },
  {
    "name": "MiniSD v1.4",
    "file": "minisd_v1.4_f16.ckpt",
    "prefix": "",
    "version": "v1",
    "defaultScale": 4,
    "deprecated": true
  },
  {
    "prefix": "modern disney style ",
    "upcast_attention": false,
    "version": "v1",
    "file": "modi_v1_f16.ckpt",
    "name": "Modern Disney (v1)",
    "text_encoder": "modi_v1_clip_vit_l14_f16.ckpt",
    "default_scale": 8
  },
  {
    "file": "nitro_v1_f16.ckpt",
    "default_scale": 8,
    "version": "v1",
    "prefix": "",
    "name": "Multi-Style (Nitro Diffusion v1)",
    "upcast_attention": false,
    "text_encoder": "nitro_v1_clip_vit_l14_f16.ckpt"
  },
  {
    "text_encoder": "mdjrny_v4_clip_vit_l14_f16.ckpt",
    "version": "v1",
    "prefix": "mdjrny-v4 style ",
    "upcast_attention": false,
    "file": "mdjrny_v4_f16.ckpt",
    "name": "Openjourney",
    "default_scale": 8
  },
  {
    "autoencoder": "vae_ft_mse_840000_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "v1",
    "upcast_attention": false,
    "name": "Outfitting Fusion Virtual Try-on (Full-Body)",
    "modifier": "double",
    "file": "ootd_vton_full_body_1.0_f16.ckpt",
    "text_encoder": "clip_vit_l14_f16.ckpt"
  },
  {
    "autoencoder": "vae_ft_mse_840000_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "v1",
    "upcast_attention": false,
    "name": "Outfitting Fusion Virtual Try-on (Upper-Body)",
    "modifier": "double",
    "file": "ootd_vton_upper_body_1.0_f16.ckpt",
    "text_encoder": "clip_vit_l14_f16.ckpt"
  },
  {
    "name": "Paper Cut (v1)",
    "upcast_attention": false,
    "version": "v1",
    "file": "papercut_v1_f16.ckpt",
    "prefix": "papercut ",
    "default_scale": 8,
    "text_encoder": "papercut_v1_clip_vit_l14_f16.ckpt"
  },
  {
    "name": "PixArt Sigma XL 1K",
    "file": "pixart_sigma_xl_2_1024_ms_f16.ckpt",
    "prefix": "",
    "version": "pixart",
    "defaultScale": 16,
    "textEncoder": "t5_xxl_encoder_q6p.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt"
  },
  {
    "name": "PixArt Sigma XL 1K (8-bit)",
    "file": "pixart_sigma_xl_2_1024_ms_q8p.ckpt",
    "prefix": "",
    "version": "pixart",
    "defaultScale": 16,
    "textEncoder": "t5_xxl_encoder_q6p.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt"
  },
  {
    "name": "PixArt Sigma XL 512",
    "file": "pixart_sigma_xl_2_512_ms_f16.ckpt",
    "prefix": "",
    "version": "pixart",
    "defaultScale": 8,
    "textEncoder": "t5_xxl_encoder_q6p.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt"
  },
  {
    "name": "PixArt Sigma XL 512 (8-bit)",
    "file": "pixart_sigma_xl_2_512_ms_q8p.ckpt",
    "prefix": "",
    "version": "pixart",
    "defaultScale": 8,
    "textEncoder": "t5_xxl_encoder_q6p.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "PixelWave 10",
    "modifier": "none",
    "file": "pixelwave_10_f16.ckpt",
    "clip_encoder": "pixelwave_10_clip_vit_l14_f16.ckpt",
    "text_encoder": "pixelwave_10_open_clip_vit_bigg14_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "PixelWave 10 (8-bit)",
    "modifier": "none",
    "file": "pixelwave_10_q6p_q8p.ckpt",
    "clip_encoder": "pixelwave_10_clip_vit_l14_f16.ckpt",
    "text_encoder": "pixelwave_10_open_clip_vit_bigg14_f16.ckpt"
  },
  {
    "name": "PixelWave FLUX.1 Schnell 04",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "pixelwave_flux_1_schnell_04_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "PixelWave FLUX.1 Schnell 04 (5-bit, SVDQuant)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "pixelwave_flux_1_schnell_04_q5p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "builtin_lora": true
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "PixelWave v9",
    "modifier": "none",
    "text_encoder": "pixelwave_v9_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "pixelwave_v9_clip_vit_l14_f16.ckpt",
    "file": "pixelwave_v9_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "PixelWave v9 (8-bit)",
    "modifier": "none",
    "text_encoder": "pixelwave_v9_open_clip_vit_bigg14_f16.ckpt",
    "clip_encoder": "pixelwave_v9_clip_vit_l14_f16.ckpt",
    "file": "pixelwave_v9_q6p_q8p.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Playground v2",
    "file": "playground_v2_f16.ckpt",
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "modifier": "none"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Playground v2 (8-bit)",
    "file": "playground_v2_q6p_q8p.ckpt",
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "modifier": "none"
  },
  {
    "conditioning": "noise",
    "latents_std": [
      8.4927,
      5.9022,
      6.5498,
      5.2299
    ],
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "objective": {
      "edm": {
        "sigma_data": 0.5
      }
    },
    "prefix": "",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Playground v2.5",
    "file": "playground_v2.5_f16.ckpt",
    "noise_discretization": {
      "edm": {
        "_0": {
          "sigma_min": 0.002,
          "sigma_data": 0.5,
          "sigma_max": 80
        }
      }
    },
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "latents_mean": [
      -1.6574,
      1.886,
      -1.383,
      2.5155
    ],
    "latents_scaling_factor": 0.5
  },
  {
    "conditioning": "noise",
    "latents_std": [
      8.4927,
      5.9022,
      6.5498,
      5.2299
    ],
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "objective": {
      "edm": {
        "sigma_data": 0.5
      }
    },
    "prefix": "",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Playground v2.5 (8-bit)",
    "file": "playground_v2.5_q6p_q8p.ckpt",
    "noise_discretization": {
      "edm": {
        "_0": {
          "sigma_min": 0.002,
          "sigma_data": 0.5,
          "sigma_max": 80
        }
      }
    },
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "latents_mean": [
      -1.6574,
      1.886,
      -1.383,
      2.5155
    ],
    "latents_scaling_factor": 0.5
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Proteus v0.3",
    "modifier": "none",
    "clip_encoder": "proteus_v0.3_clip_vit_l14_f16.ckpt",
    "file": "proteus_v0.3_f16.ckpt",
    "text_encoder": "proteus_v0.3_open_clip_vit_bigg14_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "Proteus v0.3 (8-bit)",
    "modifier": "none",
    "clip_encoder": "proteus_v0.3_clip_vit_l14_f16.ckpt",
    "file": "proteus_v0.3_q6p_q8p.ckpt",
    "text_encoder": "proteus_v0.3_open_clip_vit_bigg14_f16.ckpt"
  },
  {
    "name": "Qwen Image 1.0",
    "file": "qwen_image_1.0_q8p.ckpt",
    "prefix": "",
    "version": "qwen_image",
    "defaultScale": 16,
    "textEncoder": "qwen_2.5_vl_7b_q8p.ckpt",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "hiresFixScale": 24,
    "note": "[Qwen Image](https://huggingface.co/Qwen/Qwen-Image) is a state-of-the-art open-source image generation model known for its exceptional text layout and prompt adherence across a wide range of styles, including photorealistic, cartoon, and artistic. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Qwen Image 1.0 (6-bit)",
    "file": "qwen_image_1.0_q6p.ckpt",
    "prefix": "",
    "version": "qwen_image",
    "defaultScale": 16,
    "textEncoder": "qwen_2.5_vl_7b_q8p.ckpt",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "hiresFixScale": 24,
    "note": "[Qwen Image](https://huggingface.co/Qwen/Qwen-Image) is a state-of-the-art open-source image generation model known for its exceptional text layout and prompt adherence across a wide range of styles, including photorealistic, cartoon, and artistic. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Qwen Image 1.0 (BF16, 6-bit)",
    "file": "qwen_image_1.0_bf16_q6p.ckpt",
    "prefix": "",
    "version": "qwen_image",
    "defaultScale": 16,
    "textEncoder": "qwen_2.5_vl_7b_q8p.ckpt",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "hiresFixScale": 24,
    "isBf16": true,
    "note": "[Qwen Image](https://huggingface.co/Qwen/Qwen-Image) is a state-of-the-art open-source image generation model known for its exceptional text layout and prompt adherence across a wide range of styles, including photorealistic, cartoon, and artistic. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended. The BF16 version is only compatible with macOS 15, iOS 18 and above.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Qwen Image 1.0 (BF16, Exact)",
    "version": "qwen_image",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "qwen_image_1.0_bf16.ckpt",
    "upcast_attention": false,
    "text_encoder": "qwen_2.5_vl_7b_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "is_bf16": true,
    "note": "[Qwen Image](https://huggingface.co/Qwen/Qwen-Image) is a state-of-the-art open-source image generation model known for its exceptional text layout and prompt adherence across a wide range of styles, including photorealistic, cartoon, and artistic. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended. The BF16 version is only compatible with macOS 15, iOS 18 and above."
  },
  {
    "name": "Qwen Image 1.0 (BF16)",
    "file": "qwen_image_1.0_bf16_q8p.ckpt",
    "prefix": "",
    "version": "qwen_image",
    "defaultScale": 16,
    "textEncoder": "qwen_2.5_vl_7b_q8p.ckpt",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "hiresFixScale": 24,
    "isBf16": true,
    "note": "[Qwen Image](https://huggingface.co/Qwen/Qwen-Image) is a state-of-the-art open-source image generation model known for its exceptional text layout and prompt adherence across a wide range of styles, including photorealistic, cartoon, and artistic. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended. The BF16 version is only compatible with macOS 15, iOS 18 and above.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Qwen Image 1.0 (Exact)",
    "version": "qwen_image",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "qwen_image_1.0_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "qwen_2.5_vl_7b_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "note": "[Qwen Image](https://huggingface.co/Qwen/Qwen-Image) is a state-of-the-art open-source image generation model known for its exceptional text layout and prompt adherence across a wide range of styles, including photorealistic, cartoon, and artistic. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended."
  },
  {
    "name": "Qwen Image Edit 1.0",
    "file": "qwen_image_edit_1.0_q8p.ckpt",
    "prefix": "",
    "version": "qwen_image",
    "defaultScale": 16,
    "textEncoder": "qwen_2.5_vl_7b_q8p.ckpt",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "modifier": "kontext",
    "clipEncoder": "qwen_2.5_vl_7b_vit_f16.ckpt",
    "hiresFixScale": 24,
    "note": "[Qwen Image Edit](https://huggingface.co/Qwen/Qwen-Image-Edit) is a state-of-the-art open-source image edit model excels at image edit tasks such as background alternation, style transfer, object removal etc. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Qwen Image Edit 1.0 (6-bit)",
    "file": "qwen_image_edit_1.0_q6p.ckpt",
    "prefix": "",
    "version": "qwen_image",
    "defaultScale": 16,
    "textEncoder": "qwen_2.5_vl_7b_q8p.ckpt",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "modifier": "kontext",
    "clipEncoder": "qwen_2.5_vl_7b_vit_f16.ckpt",
    "hiresFixScale": 24,
    "note": "[Qwen Image Edit](https://huggingface.co/Qwen/Qwen-Image-Edit) is a state-of-the-art open-source image edit model excels at image edit tasks such as background alternation, style transfer, object removal etc. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Qwen Image Edit 2509",
    "file": "qwen_image_edit_2509_q8p.ckpt",
    "prefix": "",
    "version": "qwen_image",
    "defaultScale": 16,
    "textEncoder": "qwen_2.5_vl_7b_q8p.ckpt",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "modifier": "qwenimageEditPlus",
    "clipEncoder": "qwen_2.5_vl_7b_vit_f16.ckpt",
    "hiresFixScale": 24,
    "note": "[Qwen Image Edit 2509](https://huggingface.co/Qwen/Qwen-Image-Edit-2509) is a state-of-the-art open-source image edit model excels at image edit tasks such as background alternation, style transfer, object removal etc. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended. This is an update in Sep, 2025.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Qwen Image Edit 2509 (6-bit)",
    "file": "qwen_image_edit_2509_q6p.ckpt",
    "prefix": "",
    "version": "qwen_image",
    "defaultScale": 16,
    "textEncoder": "qwen_2.5_vl_7b_q8p.ckpt",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "modifier": "qwenimageEditPlus",
    "clipEncoder": "qwen_2.5_vl_7b_vit_f16.ckpt",
    "hiresFixScale": 24,
    "note": "[Qwen Image Edit 2509](https://huggingface.co/Qwen/Qwen-Image-2509) is a state-of-the-art open-source image edit model excels at image edit tasks such as background alternation, style transfer, object removal etc. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended. This is an update in Sep, 2025.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Qwen Image Edit 2509 (BF16, 6-bit)",
    "file": "qwen_image_edit_2509_bf16_q6p.ckpt",
    "prefix": "",
    "version": "qwen_image",
    "defaultScale": 16,
    "textEncoder": "qwen_2.5_vl_7b_q8p.ckpt",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "modifier": "qwenimageEditPlus",
    "clipEncoder": "qwen_2.5_vl_7b_vit_f16.ckpt",
    "hiresFixScale": 24,
    "isBf16": true,
    "note": "[Qwen Image Edit 2509](https://huggingface.co/Qwen/Qwen-Image-2509) is a state-of-the-art open-source image edit model excels at image edit tasks such as background alternation, style transfer, object removal etc. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended. This is an update in Sep, 2025. The BF16 version is only compatible with macOS 15, iOS 18 and above.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Qwen Image Edit 2509 (BF16, Exact)",
    "version": "qwen_image",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "qwen_image_edit_2509_bf16.ckpt",
    "upcast_attention": false,
    "modifier": "qwenimage_edit_plus",
    "text_encoder": "qwen_2.5_vl_7b_f16.ckpt",
    "clip_encoder": "qwen_2.5_vl_7b_vit_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "is_bf16": true,
    "note": "[Qwen Image Edit 2509](https://huggingface.co/Qwen/Qwen-Image-2509) is a state-of-the-art open-source image edit model excels at image edit tasks such as background alternation, style transfer, object removal etc. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended. This is an update in Sep, 2025. The BF16 version is only compatible with macOS 15, iOS 18 and above."
  },
  {
    "name": "Qwen Image Edit 2509 (BF16)",
    "file": "qwen_image_edit_2509_bf16_q8p.ckpt",
    "prefix": "",
    "version": "qwen_image",
    "defaultScale": 16,
    "textEncoder": "qwen_2.5_vl_7b_q8p.ckpt",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "modifier": "qwenimageEditPlus",
    "clipEncoder": "qwen_2.5_vl_7b_vit_f16.ckpt",
    "hiresFixScale": 24,
    "isBf16": true,
    "note": "[Qwen Image Edit 2509](https://huggingface.co/Qwen/Qwen-Image-Edit-2509) is a state-of-the-art open-source image edit model excels at image edit tasks such as background alternation, style transfer, object removal etc. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended. This is an update in Sep, 2025. The BF16 version is only compatible with macOS 15, iOS 18 and above.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Qwen Image Edit 2509 (Exact)",
    "version": "qwen_image",
    "autoencoder": "qwen_image_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "qwen_image_edit_2509_f16.ckpt",
    "upcast_attention": false,
    "modifier": "qwenimage_edit_plus",
    "text_encoder": "qwen_2.5_vl_7b_f16.ckpt",
    "clip_encoder": "qwen_2.5_vl_7b_vit_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "note": "[Qwen Image Edit 2509](https://huggingface.co/Qwen/Qwen-Image-2509) is a state-of-the-art open-source image edit model excels at image edit tasks such as background alternation, style transfer, object removal etc. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 30–50 sampling steps recommended. This is an update in Sep, 2025."
  },
  {
    "name": "RayFLUX v3.0",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "rayflux_v3.0_aio_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "name": "RayFLUX v3.0 (5-bit)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "rayflux_v3.0_aio_q5p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "guidance_embed": true,
    "padded_text_encoding_length": 512,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "tea_cache_coefficients": [
      498.651651,
      -283.781631,
      55.8554382,
      -3.82021401,
      0.264230861
    ]
  },
  {
    "version": "v1",
    "default_scale": 12,
    "prefix": "",
    "name": "Realistic Vision v5.1",
    "upcast_attention": false,
    "text_encoder": "realistic_vision_v5.1_clip_vit_l14_f16.ckpt",
    "file": "realistic_vision_v5.1_f16.ckpt",
    "modifier": "none"
  },
  {
    "version": "v1",
    "default_scale": 12,
    "prefix": "",
    "name": "Realistic Vision v5.1 (8-bit)",
    "upcast_attention": false,
    "text_encoder": "realistic_vision_v5.1_clip_vit_l14_f16.ckpt",
    "file": "realistic_vision_v5.1_q6p_q8p.ckpt",
    "modifier": "none"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "RealVisXL v4.0",
    "modifier": "none",
    "clip_encoder": "realvisxl_v4.0_clip_vit_l14_f16.ckpt",
    "text_encoder": "realvisxl_v4.0_open_clip_vit_bigg14_f16.ckpt",
    "file": "realvisxl_v4.0_f16.ckpt"
  },
  {
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "default_scale": 16,
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "name": "RealVisXL v4.0 (8-bit)",
    "modifier": "none",
    "clip_encoder": "realvisxl_v4.0_clip_vit_l14_f16.ckpt",
    "text_encoder": "realvisxl_v4.0_open_clip_vit_bigg14_f16.ckpt",
    "file": "realvisxl_v4.0_q6p_q8p.ckpt"
  },
  {
    "version": "v1",
    "default_scale": 8,
    "prefix": "",
    "name": "Rev Animated v1.22",
    "upcast_attention": false,
    "modifier": "none",
    "text_encoder": "rev_animated_v1.22_clip_vit_l14_f16.ckpt",
    "file": "rev_animated_v1.22_f16.ckpt"
  },
  {
    "version": "v1",
    "default_scale": 8,
    "prefix": "",
    "name": "Rev Animated v1.22 (8-bit)",
    "upcast_attention": false,
    "modifier": "none",
    "text_encoder": "rev_animated_v1.22_clip_vit_l14_f16.ckpt",
    "file": "rev_animated_v1.22_q6p_q8p.ckpt"
  },
  {
    "version": "v1",
    "name": "SamDoesArt (v3)",
    "upcast_attention": false,
    "text_encoder": "samdoesart_v3_clip_vit_l14_f16.ckpt",
    "default_scale": 8,
    "prefix": "samdoesart ",
    "file": "samdoesart_v3_f16.ckpt"
  },
  {
    "name": "schnellMODE v3.3",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "schnellmode_v3.3_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "schnellMODE v3.3 (5-bit, SVDQuant)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "schnellmode_v3.3_q5p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "builtin_lora": true
  },
  {
    "name": "SD3 Large 3.5",
    "version": "sd3_large",
    "autoencoder": "sd3_vae_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_q6p.ckpt",
    "prefix": "",
    "default_scale": 16,
    "file": "sd3_large_3.5_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "SD3 Large 3.5 (6-bit)",
    "version": "sd3_large",
    "autoencoder": "sd3_vae_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_q6p.ckpt",
    "prefix": "",
    "default_scale": 16,
    "file": "sd3_large_3.5_q6p.ckpt",
    "upcast_attention": false,
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "SD3 Large 3.5 (Exact)",
    "version": "sd3_large",
    "autoencoder": "sd3_vae_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_q6p.ckpt",
    "prefix": "",
    "default_scale": 16,
    "file": "sd3_large_3.5_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "SD3 Large Turbo 3.5",
    "version": "sd3_large",
    "autoencoder": "sd3_vae_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_q6p.ckpt",
    "prefix": "",
    "default_scale": 16,
    "file": "sd3_large_turbo_3.5_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "is_consistency_model": true
  },
  {
    "name": "SD3 Large Turbo 3.5 (6-bit)",
    "version": "sd3_large",
    "autoencoder": "sd3_vae_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_q6p.ckpt",
    "prefix": "",
    "default_scale": 16,
    "file": "sd3_large_turbo_3.5_q6p.ckpt",
    "upcast_attention": false,
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "is_consistency_model": true
  },
  {
    "name": "SD3 Medium",
    "version": "sd3",
    "autoencoder": "sd3_vae_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_q6p.ckpt",
    "prefix": "",
    "default_scale": 16,
    "file": "sd3_medium_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "SD3 Medium (8-bit)",
    "version": "sd3",
    "autoencoder": "sd3_vae_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_q6p.ckpt",
    "prefix": "",
    "default_scale": 16,
    "file": "sd3_medium_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "SD3 Medium 3.5",
    "version": "sd3",
    "autoencoder": "sd3_vae_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_q6p.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 32,
    "file": "sd3_medium_3.5_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "mmdit": {
      "qk_norm": true,
      "dual_attention_layers": [
        0,
        1,
        2,
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12
      ]
    }
  },
  {
    "name": "SD3 Medium 3.5 (8-bit)",
    "version": "sd3",
    "autoencoder": "sd3_vae_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "t5_encoder": "t5_xxl_encoder_q6p.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 32,
    "file": "sd3_medium_3.5_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "mmdit": {
      "qk_norm": true,
      "dual_attention_layers": [
        0,
        1,
        2,
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12
      ]
    }
  },
  {
    "name": "SDXL Base (v0.9)",
    "file": "sd_xl_base_0.9_f16.ckpt",
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "defaultScale": 16,
    "textEncoder": "open_clip_vit_bigg14_f16.ckpt",
    "autoencoder": "sdxl_vae_f16.ckpt",
    "deprecated": true,
    "clipEncoder": "clip_vit_l14_f16.ckpt"
  },
  {
    "name": "SDXL Base (v1.0)",
    "file": "sd_xl_base_1.0_f16.ckpt",
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "defaultScale": 16,
    "textEncoder": "open_clip_vit_bigg14_f16.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "clipEncoder": "clip_vit_l14_f16.ckpt"
  },
  {
    "name": "SDXL Base v1.0 (8-bit)",
    "file": "sd_xl_base_1.0_q6p_q8p.ckpt",
    "prefix": "",
    "version": "sdxl_base_v0.9",
    "defaultScale": 16,
    "textEncoder": "open_clip_vit_bigg14_f16.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "clipEncoder": "clip_vit_l14_f16.ckpt"
  },
  {
    "name": "SDXL Refiner (v0.9)",
    "file": "sd_xl_refiner_0.9_f16.ckpt",
    "prefix": "",
    "version": "sdxl_refiner_v0.9",
    "defaultScale": 16,
    "textEncoder": "open_clip_vit_bigg14_f16.ckpt",
    "autoencoder": "sdxl_vae_f16.ckpt",
    "deprecated": true,
    "clipEncoder": "clip_vit_l14_f16.ckpt"
  },
  {
    "name": "SDXL Refiner (v1.0)",
    "file": "sd_xl_refiner_1.0_f16.ckpt",
    "prefix": "",
    "version": "sdxl_refiner_v0.9",
    "defaultScale": 16,
    "textEncoder": "open_clip_vit_bigg14_f16.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "clipEncoder": "clip_vit_l14_f16.ckpt"
  },
  {
    "name": "SDXL Refiner v1.0 (8-bit)",
    "file": "sd_xl_refiner_1.0_q6p_q8p.ckpt",
    "prefix": "",
    "version": "sdxl_refiner_v0.9",
    "defaultScale": 16,
    "textEncoder": "open_clip_vit_bigg14_f16.ckpt",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "clipEncoder": "clip_vit_l14_f16.ckpt"
  },
  {
    "name": "SDXL Turbo",
    "prefix": "",
    "default_scale": 8,
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "file": "sd_xl_turbo_f16.ckpt",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "is_consistency_model": true
  },
  {
    "name": "SDXL Turbo (8-bit)",
    "prefix": "",
    "default_scale": 8,
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "file": "sd_xl_turbo_q6p_q8p.ckpt",
    "version": "sdxl_base_v0.9",
    "upcast_attention": false,
    "is_consistency_model": true
  },
  {
    "text_encoder": "seek_art_mega_v1_clip_vit_l14_f16.ckpt",
    "upcast_attention": false,
    "name": "seek.art MEGA (v1)",
    "version": "v1",
    "file": "seek_art_mega_v1_f16.ckpt",
    "default_scale": 10,
    "prefix": ""
  },
  {
    "name": "Shuttle Jaguar",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "shuttle_jaguar_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "Shuttle Jaguar (5-bit, SVDQuant)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "shuttle_jaguar_q5p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "builtin_lora": true
  },
  {
    "name": "Shuttle v3.1 Aesthetic",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "shuttle_v3.1_aesthetic_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    }
  },
  {
    "name": "Shuttle v3.1 Aesthetic (5-bit, SVDQuant)",
    "version": "flux1",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "shuttle_v3.1_aesthetic_q5p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "t5_xxl_encoder_q6p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": true,
    "is_consistency_model": true,
    "padded_text_encoding_length": 256,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "builtin_lora": true
  },
  {
    "name": "SkyReels v1 Hunyuan I2V 544p",
    "version": "hunyuan_video",
    "autoencoder": "hunyuan_video_vae_f16.ckpt",
    "prefix": "FPS-24, ",
    "default_scale": 12,
    "hires_fix_scale": 24,
    "file": "skyreels_v1_hunyuan_i2v_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "llava_llama_3_8b_v1.1_q8p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": false,
    "modifier": "inpainting",
    "guidance_embed": true,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      733.226126,
      -401.131952,
      67.5869174,
      -3.149878,
      0.0961237896
    ],
    "note": "[SkyReels v1](https://huggingface.co/collections/Skywork/skyreels-v1-67b34676ff65b4ec02d16307) is a significant fine-tune of Hunyuan Video, Tencent’s state-of-the-art text-to-video model. SkyReels v1 I2V is an image-to-video (img2vid) fine-tune that generates video clips up to 4 seconds long from a single image. The model supports up to 97 frames, with a recommended shift value of 7.0 or higher. For optimal results, disable \"Speedup w/ Guidance Embed\" by setting it to 1.0 and use a Text Guidance value of 6."
  },
  {
    "name": "SkyReels v1 Hunyuan I2V 544p (5-bit, SVDQuant)",
    "version": "hunyuan_video",
    "autoencoder": "hunyuan_video_vae_f16.ckpt",
    "prefix": "FPS-24, ",
    "default_scale": 12,
    "hires_fix_scale": 24,
    "file": "skyreels_v1_hunyuan_i2v_q5p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "llava_llama_3_8b_v1.1_q8p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": false,
    "modifier": "inpainting",
    "guidance_embed": true,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      733.226126,
      -401.131952,
      67.5869174,
      -3.149878,
      0.0961237896
    ],
    "note": "[SkyReels v1](https://huggingface.co/collections/Skywork/skyreels-v1-67b34676ff65b4ec02d16307) is a significant fine-tune of Hunyuan Video, Tencent’s state-of-the-art text-to-video model. SkyReels v1 I2V is an image-to-video (img2vid) fine-tune that generates video clips up to 4 seconds long from a single image. The model supports up to 97 frames, with a recommended shift value of 7.0 or higher. For optimal results, disable \"Speedup w/ Guidance Embed\" by setting it to 1.0 and use a Text Guidance value of 6.",
    "builtin_lora": true
  },
  {
    "name": "SkyReels v1 Hunyuan T2V 544p",
    "version": "hunyuan_video",
    "autoencoder": "hunyuan_video_vae_f16.ckpt",
    "prefix": "FPS-24, ",
    "default_scale": 12,
    "hires_fix_scale": 24,
    "file": "skyreels_v1_hunyuan_t2v_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "llava_llama_3_8b_v1.1_q8p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": false,
    "guidance_embed": true,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      733.226126,
      -401.131952,
      67.5869174,
      -3.149878,
      0.0961237896
    ],
    "note": "[SkyReels v1](https://huggingface.co/collections/Skywork/skyreels-v1-67b34676ff65b4ec02d16307) is a significant fine-tune of Hunyuan Video, Tencent’s state-of-the-art text-to-video model. It can generate video clips up to 4 seconds long at a recommended resolution of 960×544. The model supports up to 97 frames, with a recommended shift value of 7.0 or higher. For optimal results, disable \"Speedup w/ Guidance Embed\" by setting it to 1.0 and use a Text Guidance value of 6."
  },
  {
    "name": "SkyReels v1 Hunyuan T2V 544p (5-bit, SVDQuant)",
    "version": "hunyuan_video",
    "autoencoder": "hunyuan_video_vae_f16.ckpt",
    "prefix": "FPS-24, ",
    "default_scale": 12,
    "hires_fix_scale": 24,
    "file": "skyreels_v1_hunyuan_t2v_q5p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "llava_llama_3_8b_v1.1_q8p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "high_precision_autoencoder": false,
    "guidance_embed": true,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      733.226126,
      -401.131952,
      67.5869174,
      -3.149878,
      0.0961237896
    ],
    "note": "[SkyReels v1](https://huggingface.co/collections/Skywork/skyreels-v1-67b34676ff65b4ec02d16307) is a significant fine-tune of Hunyuan Video, Tencent’s state-of-the-art text-to-video model. It can generate video clips up to 4 seconds long at a recommended resolution of 960×544. The model supports up to 97 frames, with a recommended shift value of 7.0 or higher. For optimal results, disable \"Speedup w/ Guidance Embed\" by setting it to 1.0 and use a Text Guidance value of 6.",
    "builtin_lora": true
  },
  {
    "name": "SkyReels v2 I2V 1.3B 540p",
    "version": "wan_v2.1_1.3b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 8,
    "hires_fix_scale": 12,
    "file": "skyreels_v2_i2v_1.3b_540p_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "note": "[SkyReels v2 I2V 1.3B 540p](https://huggingface.co/Skywork/SkyReels-V2-I2V-1.3B-540P) is a image-to-video model developed by Skywork AI. It can generate video clips of up to 4 seconds in length. The model is trained at 544x960, supports up to 97 frames, with a recommended shift value of 5.0. As a derivative of Wan 2.1 (1.3B), it performs best with similar generation settings."
  },
  {
    "name": "SkyReels v2 I2V 1.3B 540p (8-bit)",
    "version": "wan_v2.1_1.3b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 8,
    "hires_fix_scale": 12,
    "file": "skyreels_v2_i2v_1.3b_540p_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "note": "[SkyReels v2 I2V 1.3B 540p](https://huggingface.co/Skywork/SkyReels-V2-I2V-1.3B-540P) is a image-to-video model developed by Skywork AI. It can generate video clips of up to 4 seconds in length. The model is trained at 544x960, supports up to 97 frames, with a recommended shift value of 5.0. As a derivative of Wan 2.1 (1.3B), it performs best with similar generation settings."
  },
  {
    "name": "SkyReels v2 I2V 14B 540p",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "skyreels_v2_i2v_14b_540p_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "note": "[SkyReels v2 I2V 14B 540p](https://huggingface.co/Skywork/SkyReels-V2-I2V-14B-540P) is a image-to-video model developed by Skywork AI. It can generate video clips of up to 4 seconds in length. The model is trained at 544x960, supports up to 97 frames, with a recommended shift value of 5.0. As a derivative of Wan 2.1 (14B), it performs best with similar generation settings."
  },
  {
    "name": "SkyReels v2 I2V 14B 540p (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "skyreels_v2_i2v_14b_540p_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "note": "[SkyReels v2 I2V 14B 540p](https://huggingface.co/Skywork/SkyReels-V2-I2V-14B-540P) is a image-to-video model developed by Skywork AI. It can generate video clips of up to 4 seconds in length. The model is trained at 544x960, supports up to 97 frames, with a recommended shift value of 5.0. As a derivative of Wan 2.1 (14B), it performs best with similar generation settings.",
    "builtin_lora": true
  },
  {
    "name": "SkyReels v2 I2V 14B 720p",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "skyreels_v2_i2v_14b_720p_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "note": "[SkyReels v2 I2V 14B 720p](https://huggingface.co/Skywork/SkyReels-V2-I2V-14B-720P) is a image-to-video model developed by Skywork AI. It can generate video clips of up to 5 seconds in length. The model is trained at 720x1280, supports up to 121 frames, with a recommended shift value of 5.0. As a derivative of Wan 2.1 (14B), it performs best with similar generation settings."
  },
  {
    "name": "SkyReels v2 I2V 14B 720p (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "skyreels_v2_i2v_14b_720p_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "note": "[SkyReels v2 I2V 14B 720p](https://huggingface.co/Skywork/SkyReels-V2-I2V-14B-720P) is a image-to-video model developed by Skywork AI. It can generate video clips of up to 5 seconds in length. The model is trained at 720x1280, supports up to 121 frames, with a recommended shift value of 5.0. As a derivative of Wan 2.1 (14B), it performs best with similar generation settings.",
    "builtin_lora": true
  },
  {
    "name": "SkyReels v2 T2V 14B 540p",
    "version": "wan_v2.1_14b",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "skyreels_v2_t2v_14b_540p_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "note": "[SkyReels v2 T2V 14B 540p](https://huggingface.co/Skywork/SkyReels-V2-T2V-14B-540P) is a text-to-video model developed by Skywork AI. It can generate video clips of up to 4 seconds in length. The model is trained at 544x960, supports up to 97 frames, with a recommended shift value of 5.0. As a derivative of Wan 2.1 (14B), it performs best with similar generation settings."
  },
  {
    "name": "SkyReels v2 T2V 14B 540p (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "skyreels_v2_t2v_14b_540p_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "note": "[SkyReels v2 T2V 14B 540p](https://huggingface.co/Skywork/SkyReels-V2-T2V-14B-540P) is a text-to-video model developed by Skywork AI. It can generate video clips of up to 4 seconds in length. The model is trained at 544x960, supports up to 97 frames, with a recommended shift value of 5.0. As a derivative of Wan 2.1 (14B), it performs best with similar generation settings.",
    "builtin_lora": true
  },
  {
    "name": "SkyReels v2 T2V 14B 720p",
    "version": "wan_v2.1_14b",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "skyreels_v2_t2v_14b_720p_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "note": "[SkyReels v2 T2V 14B 720p](https://huggingface.co/Skywork/SkyReels-V2-T2V-14B-720P) is a text-to-video model developed by Skywork AI. It can generate video clips of up to 5 seconds in length. The model is trained at 720x1280, supports up to 121 frames, with a recommended shift value of 5.0. As a derivative of Wan 2.1 (14B), it performs best with similar generation settings."
  },
  {
    "name": "SkyReels v2 T2V 14B 720p (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "skyreels_v2_t2v_14b_720p_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 24,
    "tea_cache_coefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "note": "[SkyReels v2 T2V 14B 720p](https://huggingface.co/Skywork/SkyReels-V2-T2V-14B-720P) is a text-to-video model developed by Skywork AI. It can generate video clips of up to 5 seconds in length. The model is trained at 720x1280, supports up to 121 frames, with a recommended shift value of 5.0. As a derivative of Wan 2.1 (14B), it performs best with similar generation settings.",
    "builtin_lora": true
  },
  {
    "name": "Spider-Verse (v1)",
    "version": "v1",
    "text_encoder": "spiderverse_v1_clip_vit_l14_f16.ckpt",
    "upcast_attention": false,
    "default_scale": 8,
    "file": "spiderverse_v1_f16.ckpt",
    "prefix": "spiderverse style "
  },
  {
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "prefix": "",
    "file": "ssd_1b_f16.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "version": "ssd_1b",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "upcast_attention": false,
    "name": "SSD 1B (Segmind SDXL)",
    "default_scale": 16
  },
  {
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "prefix": "",
    "file": "ssd_1b_q6p_q8p.ckpt",
    "clip_encoder": "clip_vit_l14_f16.ckpt",
    "version": "ssd_1b",
    "autoencoder": "sdxl_vae_v1.0_f16.ckpt",
    "upcast_attention": false,
    "name": "SSD 1B (Segmind SDXL) (8-bit)",
    "default_scale": 16
  },
  {
    "name": "Stable Cascade (Würstchen v3.0)",
    "stage_models": [
      "wurstchen_3.0_stage_b_f16.ckpt"
    ],
    "version": "wurstchen_v3.0_stage_c",
    "default_scale": 16,
    "file": "wurstchen_3.0_stage_c_f32_f16.ckpt",
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "autoencoder": "wurstchen_3.0_stage_a_hq_f16.ckpt",
    "upcast_attention": false,
    "prefix": ""
  },
  {
    "name": "Stable Cascade (Würstchen v3.0) (8-bit)",
    "stage_models": [
      "wurstchen_3.0_stage_b_q6p_q8p.ckpt"
    ],
    "version": "wurstchen_v3.0_stage_c",
    "default_scale": 16,
    "file": "wurstchen_3.0_stage_c_f32_q6p_q8p.ckpt",
    "text_encoder": "open_clip_vit_bigg14_f16.ckpt",
    "autoencoder": "wurstchen_3.0_stage_a_hq_f16.ckpt",
    "upcast_attention": false,
    "prefix": ""
  },
  {
    "name": "Stable Diffusion v1.4",
    "file": "sd_v1.4_f16.ckpt",
    "prefix": "",
    "version": "v1",
    "deprecated": true
  },
  {
    "name": "Stable Diffusion v1.5",
    "file": "sd_v1.5_f16.ckpt",
    "prefix": "",
    "version": "v1"
  },
  {
    "name": "Stable Diffusion v1.5 Inpainting",
    "file": "sd_v1.5_inpainting_f16.ckpt",
    "prefix": "",
    "version": "v1",
    "modifier": "inpainting"
  },
  {
    "name": "Stable Diffusion v2.0",
    "file": "sd_v2.0_f16.ckpt",
    "prefix": "",
    "version": "v2",
    "textEncoder": "open_clip_vit_h14_f16.ckpt",
    "deprecated": true
  },
  {
    "name": "Stable Diffusion v2.0 768-v",
    "file": "sd_v2.0_768_v_f16.ckpt",
    "prefix": "",
    "version": "v2",
    "defaultScale": 12,
    "textEncoder": "open_clip_vit_h14_f16.ckpt",
    "deprecated": true
  },
  {
    "name": "Stable Diffusion v2.0 Depth",
    "file": "sd_v2.0_depth_f16.ckpt",
    "prefix": "",
    "version": "v2",
    "textEncoder": "open_clip_vit_h14_f16.ckpt",
    "modifier": "depth"
  },
  {
    "name": "Stable Diffusion v2.0 Inpainting",
    "file": "sd_v2.0_inpainting_f16.ckpt",
    "prefix": "",
    "version": "v2",
    "textEncoder": "open_clip_vit_h14_f16.ckpt",
    "modifier": "inpainting"
  },
  {
    "name": "Stable Diffusion v2.1",
    "file": "sd_v2.1_f16.ckpt",
    "prefix": "",
    "version": "v2",
    "textEncoder": "open_clip_vit_h14_f16.ckpt"
  },
  {
    "name": "Stable Diffusion v2.1 768-v",
    "file": "sd_v2.1_768_v_f16.ckpt",
    "prefix": "",
    "version": "v2",
    "upcastAttention": true,
    "defaultScale": 12,
    "textEncoder": "open_clip_vit_h14_f16.ckpt"
  },
  {
    "file": "svd_i2v_1.0_f16.ckpt",
    "version": "svd_i2v",
    "conditioning": "noise",
    "objective": {
      "v": {}
    },
    "prefix": "",
    "default_scale": 8,
    "clip_encoder": "svd_i2v_1.0_f16.ckpt",
    "name": "Stable Video Diffusion I2V v1.0",
    "text_encoder": "open_clip_vit_h14_vision_model_f16.ckpt",
    "upcast_attention": false,
    "modifier": "none",
    "noise_discretization": {
      "edm": {
        "_0": {
          "sigma_max": 700,
          "sigma_data": 0.5,
          "sigma_min": 0.002
        }
      }
    }
  },
  {
    "file": "svd_i2v_1.0_q6p_q8p.ckpt",
    "version": "svd_i2v",
    "conditioning": "noise",
    "objective": {
      "v": {}
    },
    "prefix": "",
    "default_scale": 8,
    "clip_encoder": "svd_i2v_1.0_q6p_q8p.ckpt",
    "name": "Stable Video Diffusion I2V v1.0 (8-bit)",
    "text_encoder": "open_clip_vit_h14_vision_model_f16.ckpt",
    "upcast_attention": false,
    "modifier": "none",
    "noise_discretization": {
      "edm": {
        "_0": {
          "sigma_max": 700,
          "sigma_data": 0.5,
          "sigma_min": 0.002
        }
      }
    }
  },
  {
    "name": "Stable Video Diffusion I2V XT 1.0 (8-bit)",
    "file": "svd_i2v_xt_1.0_q6p_q8p.ckpt",
    "prefix": "",
    "version": "svd_i2v",
    "defaultScale": 8,
    "textEncoder": "open_clip_vit_h14_vision_model_f16.ckpt",
    "deprecated": true,
    "clipEncoder": "svd_i2v_xt_1.0_q6p_q8p.ckpt"
  },
  {
    "name": "Stable Video Diffusion I2V XT v1.0",
    "file": "svd_i2v_xt_1.0_f16.ckpt",
    "prefix": "",
    "version": "svd_i2v",
    "defaultScale": 8,
    "textEncoder": "open_clip_vit_h14_vision_model_f16.ckpt",
    "deprecated": true,
    "clipEncoder": "svd_i2v_xt_1.0_f16.ckpt"
  },
  {
    "file": "svd_i2v_xt_1.1_f16.ckpt",
    "version": "svd_i2v",
    "conditioning": "noise",
    "objective": {
      "v": {}
    },
    "prefix": "",
    "default_scale": 8,
    "clip_encoder": "svd_i2v_xt_1.1_f16.ckpt",
    "name": "Stable Video Diffusion I2V XT v1.1",
    "text_encoder": "open_clip_vit_h14_vision_model_f16.ckpt",
    "upcast_attention": false,
    "modifier": "none",
    "noise_discretization": {
      "edm": {
        "_0": {
          "sigma_max": 700,
          "sigma_data": 0.5,
          "sigma_min": 0.002
        }
      }
    }
  },
  {
    "file": "svd_i2v_xt_1.1_q6p_q8p.ckpt",
    "version": "svd_i2v",
    "conditioning": "noise",
    "objective": {
      "v": {}
    },
    "prefix": "",
    "default_scale": 8,
    "clip_encoder": "svd_i2v_xt_1.1_q6p_q8p.ckpt",
    "name": "Stable Video Diffusion I2V XT v1.1 (8-bit)",
    "text_encoder": "open_clip_vit_h14_vision_model_f16.ckpt",
    "upcast_attention": false,
    "modifier": "none",
    "noise_discretization": {
      "edm": {
        "_0": {
          "sigma_max": 700,
          "sigma_data": 0.5,
          "sigma_min": 0.002
        }
      }
    }
  },
  {
    "name": "Super Mario Nation (v2)",
    "default_scale": 8,
    "text_encoder": "supermarionation_v2_clip_vit_l14_f16.ckpt",
    "file": "supermarionation_v2_f16.ckpt",
    "prefix": "supermarionation ",
    "version": "v1",
    "upcast_attention": false
  },
  {
    "version": "v1",
    "prefix": "trnlgcy style ",
    "upcast_attention": false,
    "default_scale": 8,
    "text_encoder": "trnlgcy_clip_vit_l14_f16.ckpt",
    "name": "Tron Legacy",
    "file": "trnlgcy_f16.ckpt"
  },
  {
    "name": "Van Gogh Style (Lvngvncnt v2)",
    "version": "v1",
    "text_encoder": "lvngvncnt_v2_clip_vit_l14_f16.ckpt",
    "upcast_attention": false,
    "default_scale": 8,
    "file": "lvngvncnt_v2_f16.ckpt",
    "prefix": "lvngvncnt "
  },
  {
    "name": "VoxelArt (v1)",
    "upcast_attention": false,
    "version": "v1",
    "file": "voxelart_v1_f16.ckpt",
    "prefix": "voxelart ",
    "default_scale": 8,
    "text_encoder": "voxelart_v1_clip_vit_l14_f16.ckpt"
  },
  {
    "name": "Wan 2.1 1.3B Fun InP",
    "version": "wan_v2.1_1.3b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 8,
    "hires_fix_scale": 12,
    "file": "wan_2.1_1.3b_fun_inp_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      -52186.2437,
      9230.41404,
      -528.275948,
      13.6987616,
      -0.0499875664
    ],
    "note": "[Wan2.1 1.3B Fun InP](https://huggingface.co/alibaba-pai/Wan2.1-Fun-1.3B-InP) is a state-of-the-art image-to-video model developed by Alibaba PAI. It can generate video clips of up to 4 seconds in length. The model is trained at multiple resolutions, supports up to 81 frames, with a recommended shift value of 5.0. Text Guidance should be somewhere between 3 to 7. Wan2.1 is trained with Flow Matching objective, Trailing samplers would give the best result."
  },
  {
    "name": "Wan 2.1 1.3B Fun InP (8-bit)",
    "version": "wan_v2.1_1.3b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 8,
    "hires_fix_scale": 12,
    "file": "wan_2.1_1.3b_fun_inp_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      -52186.2437,
      9230.41404,
      -528.275948,
      13.6987616,
      -0.0499875664
    ],
    "note": "[Wan2.1 1.3B Fun InP](https://huggingface.co/alibaba-pai/Wan2.1-Fun-1.3B-InP) is a state-of-the-art image-to-video model developed by Alibaba PAI. It can generate video clips of up to 4 seconds in length. The model is trained at multiple resolutions, supports up to 81 frames, with a recommended shift value of 5.0. Text Guidance should be somewhere between 3 to 7. Wan2.1 is trained with Flow Matching objective, Trailing samplers would give the best result."
  },
  {
    "name": "Wan 2.1 1.3B v1.1 Fun InP",
    "version": "wan_v2.1_1.3b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 8,
    "hires_fix_scale": 12,
    "file": "wan_2.1_1.3b_v1.1_fun_inp_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      -52186.2437,
      9230.41404,
      -528.275948,
      13.6987616,
      -0.0499875664
    ],
    "note": "[Wan2.1 1.3B v1.1 Fun InP](https://huggingface.co/alibaba-pai/Wan2.1-Fun-V1.1-1.3B-InP) is a state-of-the-art image-to-video model developed by Alibaba PAI. It can generate video clips of up to 4 seconds in length. The model is trained at multiple resolutions, supports up to 81 frames, with a recommended shift value of 5.0. Text Guidance should be somewhere between 3 to 7. Wan2.1 is trained with Flow Matching objective, Trailing samplers would give the best result."
  },
  {
    "name": "Wan 2.1 1.3B v1.1 Fun InP (8-bit)",
    "version": "wan_v2.1_1.3b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 8,
    "hires_fix_scale": 12,
    "file": "wan_2.1_1.3b_v1.1_fun_inp_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      -52186.2437,
      9230.41404,
      -528.275948,
      13.6987616,
      -0.0499875664
    ],
    "note": "[Wan2.1 1.3B v1.1 Fun InP](https://huggingface.co/alibaba-pai/Wan2.1-Fun-V1.1-1.3B-InP) is a state-of-the-art image-to-video model developed by Alibaba PAI. It can generate video clips of up to 4 seconds in length. The model is trained at multiple resolutions, supports up to 81 frames, with a recommended shift value of 5.0. Text Guidance should be somewhere between 3 to 7. Wan2.1 is trained with Flow Matching objective, Trailing samplers would give the best result."
  },
  {
    "name": "Wan 2.1 14B Fun InP",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "wan_2.1_14b_fun_inp_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "note": "[Wan2.1 14B Fun InP](https://huggingface.co/alibaba-pai/Wan2.1-Fun-14B-InP) is a state-of-the-art image-to-video model developed by Alibaba PAI. It can generate video clips of up to 4 seconds in length. The model is trained at multiple resolutions, supports up to 81 frames, with a recommended shift value of 5.0. Text Guidance should be somewhere between 3 to 7. Wan2.1 is trained with Flow Matching objective, Trailing samplers would give the best result."
  },
  {
    "name": "Wan 2.1 14B Fun InP (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "wan_2.1_14b_fun_inp_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "note": "[Wan2.1 14B Fun InP](https://huggingface.co/alibaba-pai/Wan2.1-Fun-14B-InP) is a state-of-the-art image-to-video model developed by Alibaba PAI. It can generate video clips of up to 4 seconds in length. The model is trained at multiple resolutions, supports up to 81 frames, with a recommended shift value of 5.0. Text Guidance should be somewhere between 3 to 7. Wan2.1 is trained with Flow Matching objective, Trailing samplers would give the best result.",
    "builtin_lora": true
  },
  {
    "name": "Wan 2.1 14B I2V FusionX",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "wan_2.1_14b_i2v_fusionx_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "note": "[Wan2.1 14B FusionX](https://huggingface.co/vrgamedevgirl84/Wan14BT2VFusioniX) is a merged model of multiple quality and acceleration LoRAs."
  },
  {
    "name": "Wan 2.1 14B I2V FusionX (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "wan_2.1_14b_i2v_fusionx_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "note": "[Wan2.1 14B FusionX](https://huggingface.co/vrgamedevgirl84/Wan14BT2VFusioniX) is a merged model of multiple quality and acceleration LoRAs.",
    "builtin_lora": true
  },
  {
    "name": "Wan 2.1 14B T2V FusionX",
    "version": "wan_v2.1_14b",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "wan_2.1_14b_t2v_fusionx_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "note": "[Wan2.1 14B FusionX](https://huggingface.co/vrgamedevgirl84/Wan14BT2VFusioniX) is a merged model of multiple quality and acceleration LoRAs."
  },
  {
    "name": "Wan 2.1 14B T2V FusionX (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "wan_2.1_14b_t2v_fusionx_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "note": "[Wan2.1 14B FusionX](https://huggingface.co/vrgamedevgirl84/Wan14BT2VFusioniX) is a merged model of multiple quality and acceleration LoRAs.",
    "builtin_lora": true
  },
  {
    "name": "Wan 2.1 14B v1.1 Fun InP",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "wan_2.1_14b_v1.1_fun_inp_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "note": "[Wan2.1 14B v1.1 Fun InP](https://huggingface.co/alibaba-pai/Wan2.1-Fun-V1.1-14B-InP) is a state-of-the-art image-to-video model developed by Alibaba PAI. It can generate video clips of up to 4 seconds in length. The model is trained at multiple resolutions, supports up to 81 frames, with a recommended shift value of 5.0. Text Guidance should be somewhere between 3 to 7. Wan2.1 is trained with Flow Matching objective, Trailing samplers would give the best result."
  },
  {
    "name": "Wan 2.1 14B v1.1 Fun InP (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "modifier": "inpainting",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "wan_2.1_14b_v1.1_fun_inp_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "clip_encoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "note": "[Wan2.1 14B v1.1 Fun InP](https://huggingface.co/alibaba-pai/Wan2.1-Fun-V1.1-14B-InP) is a state-of-the-art image-to-video model developed by Alibaba PAI. It can generate video clips of up to 4 seconds in length. The model is trained at multiple resolutions, supports up to 81 frames, with a recommended shift value of 5.0. Text Guidance should be somewhere between 3 to 7. Wan2.1 is trained with Flow Matching objective, Trailing samplers would give the best result.",
    "builtin_lora": true
  },
  {
    "name": "Wan 2.1 I2V 14B 480p",
    "file": "wan_v2.1_14b_i2v_480p_q8p.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 8,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "modifier": "inpainting",
    "clipEncoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "hiresFixScale": 12,
    "teaCacheCoefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.1 I2V 14B 480P](https://huggingface.co/Wan-AI/Wan2.1-I2V-14B-480P) is a state-of-the-art image-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length from a given start frame. The recommended resolutions are 832×480. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.1 I2V 14B 480p (6-bit, SVDQuant)",
    "file": "wan_v2.1_14b_i2v_480p_q6p_svd.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 8,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "modifier": "inpainting",
    "clipEncoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "hiresFixScale": 12,
    "builtinLora": true,
    "teaCacheCoefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.1 I2V 14B 480P](https://huggingface.co/Wan-AI/Wan2.1-I2V-14B-480P) is a state-of-the-art image-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length from a given start frame. The recommended resolutions are 832×480. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.1 I2V 14B 720p",
    "file": "wan_v2.1_14b_i2v_720p_q8p.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "modifier": "inpainting",
    "clipEncoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "hiresFixScale": 16,
    "teaCacheCoefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.1 I2V 14B 720P](https://huggingface.co/Wan-AI/Wan2.1-I2V-14B-720P) is a state-of-the-art image-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length from a given start frame. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.1 I2V 14B 720p (6-bit, SVDQuant)",
    "file": "wan_v2.1_14b_i2v_720p_q6p_svd.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "modifier": "inpainting",
    "clipEncoder": "open_clip_xlm_roberta_large_vit_h14_f16.ckpt",
    "hiresFixScale": 16,
    "builtinLora": true,
    "teaCacheCoefficients": [
      8107.0546,
      2133.93892,
      -372.934672,
      16.6203073,
      -0.0417769401
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.1 I2V 14B 720P](https://huggingface.co/Wan-AI/Wan2.1-I2V-14B-720P) is a state-of-the-art image-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length from a given start frame. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.1 T2V 1.3B",
    "file": "wan_v2.1_1.3b_480p_f16.ckpt",
    "prefix": "",
    "version": "wan_v2.1_1.3b",
    "defaultScale": 8,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "hiresFixScale": 12,
    "teaCacheCoefficients": [
      -52186.2437,
      9230.41404,
      -528.275948,
      13.6987616,
      -0.0499875664
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.1 T2V 1.3B](https://huggingface.co/Wan-AI/Wan2.1-T2V-1.3B) is a state-of-the-art text-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 832×480. The model supports up to 81 frames, with a recommended shift value of 6.0. For best results, set Text Guidance above 5.0. Wan2.1 is trained with a Flow Matching objective, and trailing samplers will produce the best outputs.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.1 T2V 1.3B (8-bit)",
    "file": "wan_v2.1_1.3b_480p_q8p.ckpt",
    "prefix": "",
    "version": "wan_v2.1_1.3b",
    "defaultScale": 8,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "hiresFixScale": 12,
    "teaCacheCoefficients": [
      -52186.2437,
      9230.41404,
      -528.275948,
      13.6987616,
      -0.0499875664
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.1 T2V 1.3B](https://huggingface.co/Wan-AI/Wan2.1-T2V-1.3B) is a state-of-the-art text-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 832×480. The model supports up to 81 frames, with a recommended shift value of 6.0. For best results, set Text Guidance above 5.0. Wan2.1 is trained with a Flow Matching objective, and trailing samplers will produce the best outputs.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.1 T2V 14B",
    "file": "wan_v2.1_14b_720p_q8p.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "hiresFixScale": 16,
    "teaCacheCoefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.1 T2V 14B](https://huggingface.co/Wan-AI/Wan2.1-T2V-14B) is a state-of-the-art text-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 1280×720. The recommended resolutions are 832×480. The model supports up to 81 frames, with a recommended shift value of 5.0. For best results, set Text Guidance above 5.0. Wan2.1 is trained with a Flow Matching objective, and trailing samplers will produce the best outputs.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.1 T2V 14B (5-bit, SVDQuant)",
    "file": "wan_v2.1_14b_720p_q5p_svd.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "deprecated": true,
    "hiresFixScale": 16,
    "builtinLora": true,
    "teaCacheCoefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.1 T2V 14B](https://huggingface.co/Wan-AI/Wan2.1-T2V-14B) is a state-of-the-art text-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.1 T2V 14B (6-bit, SVDQuant)",
    "file": "wan_v2.1_14b_720p_q6p_svd.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "hiresFixScale": 16,
    "builtinLora": true,
    "teaCacheCoefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.1 T2V 14B](https://huggingface.co/Wan-AI/Wan2.1-T2V-14B) is a state-of-the-art text-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.2 High Noise Expert I2V A14B",
    "file": "wan_v2.2_a14b_hne_i2v_q8p.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "modifier": "inpainting",
    "hiresFixScale": 16,
    "teaCacheCoefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.2 I2V A14B](https://huggingface.co/Wan-AI/Wan2.2-I2V-A14B) is a state-of-the-art image-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length from a given start frame. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.2 High Noise Expert I2V A14B (6-bit, SVDQuant)",
    "file": "wan_v2.2_a14b_hne_i2v_q6p_svd.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "modifier": "inpainting",
    "hiresFixScale": 16,
    "builtinLora": true,
    "teaCacheCoefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.2 I2V A14B](https://huggingface.co/Wan-AI/Wan2.2-I2V-A14B) is a state-of-the-art image-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length from a given start frame. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.2 High Noise Expert T2V A14B",
    "file": "wan_v2.2_a14b_hne_t2v_q8p.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "hiresFixScale": 16,
    "teaCacheCoefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.2 T2V A14B](https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B) is a state-of-the-art text-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.2 High Noise Expert T2V A14B (6-bit, SVDQuant)",
    "file": "wan_v2.2_a14b_hne_t2v_q6p_svd.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "hiresFixScale": 16,
    "builtinLora": true,
    "teaCacheCoefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.2 T2V A14B](https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B) is a state-of-the-art text-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.2 Low Noise Expert I2V A14B",
    "file": "wan_v2.2_a14b_lne_i2v_q8p.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "modifier": "inpainting",
    "hiresFixScale": 16,
    "teaCacheCoefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.2 I2V A14B](https://huggingface.co/Wan-AI/Wan2.2-I2V-A14B) is a state-of-the-art image-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length from a given start frame. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.2 Low Noise Expert I2V A14B (6-bit, SVDQuant)",
    "file": "wan_v2.2_a14b_lne_i2v_q6p_svd.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "modifier": "inpainting",
    "hiresFixScale": 16,
    "builtinLora": true,
    "teaCacheCoefficients": [
      257151.496,
      -35422.9917,
      1402.86849,
      -13.5890334,
      0.132517977
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.2 I2V A14B](https://huggingface.co/Wan-AI/Wan2.2-I2V-A14B) is a state-of-the-art image-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length from a given start frame. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.2 Low Noise Expert T2V A14B",
    "file": "wan_v2.2_a14b_lne_t2v_q8p.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "hiresFixScale": 16,
    "teaCacheCoefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.2 T2V A14B](https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B) is a state-of-the-art text-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.2 Low Noise Expert T2V A14B (6-bit, SVDQuant)",
    "file": "wan_v2.2_a14b_lne_t2v_q6p_svd.ckpt",
    "prefix": "",
    "version": "wan_v2.1_14b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "hiresFixScale": 16,
    "builtinLora": true,
    "teaCacheCoefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "framesPerSecond": 16,
    "note": "[Wan2.2 T2V A14B](https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B) is a state-of-the-art text-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 1280×720. The model supports up to 81 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.2 TI2V 5B",
    "file": "wan_v2.2_5b_ti2v_f16.ckpt",
    "prefix": "",
    "version": "wan_v2.2_5b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.2_video_vae_f16.ckpt",
    "hiresFixScale": 16,
    "teaCacheCoefficients": null,
    "framesPerSecond": 24,
    "note": "[Wan2.2 TI2V 5B](https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B) is a state-of-the-art text-image-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 1280×720. The model supports up to 121 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan 2.2 TI2V 5B (8-bit)",
    "file": "wan_v2.2_5b_ti2v_q8p.ckpt",
    "prefix": "",
    "version": "wan_v2.2_5b",
    "defaultScale": 12,
    "textEncoder": "umt5_xxl_encoder_q8p.ckpt",
    "autoencoder": "wan_v2.2_video_vae_f16.ckpt",
    "hiresFixScale": 16,
    "teaCacheCoefficients": null,
    "framesPerSecond": 24,
    "note": "[Wan2.2 TI2V 5B](https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B) is a state-of-the-art text-image-to-video model developed by Alibaba. It can generate video clips of up to 4 seconds in length. The recommended resolutions are 1280×720. The model supports up to 121 frames, with a recommended shift value of 5.0.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Wan v2.2 High Noise Expert T2V A14B Lightning 250928",
    "version": "wan_v2.1_14b",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "wan_v2.2_a14b_hne_t2v_lightning_250928_q8p.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "note": "[Wan2.2 High Noise Expert T2V A14B Lightning 250928](https://huggingface.co/lightx2v/Wan2.2-Lightning/tree/main/Wan2.2-T2V-A14B-4steps-250928-dyno) is a fine-tune of Wan v2.2 High Noise Expert model. It can generate more dynamic motions with 4 steps and text guidance 1.0."
  },
  {
    "name": "Wan v2.2 High Noise Expert T2V A14B Lightning 250928 (6-bit, SVDQuant)",
    "version": "wan_v2.1_14b",
    "autoencoder": "wan_v2.1_video_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 12,
    "hires_fix_scale": 16,
    "file": "wan_v2.2_a14b_hne_t2v_lightning_250928_q6p_svd.ckpt",
    "upcast_attention": false,
    "text_encoder": "umt5_xxl_encoder_q8p.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "frames_per_second": 16,
    "tea_cache_coefficients": [
      -303318.725,
      49053.7029,
      -2655.30556,
      58.7365115,
      -0.315583525
    ],
    "note": "[Wan2.2 High Noise Expert T2V A14B Lightning 250928](https://huggingface.co/lightx2v/Wan2.2-Lightning/tree/main/Wan2.2-T2V-A14B-4steps-250928-dyno) is a fine-tune of Wan v2.2 High Noise Expert model. It can generate more dynamic motions with 4 steps and text guidance 1.0.",
    "builtin_lora": true
  },
  {
    "name": "Z Image Turbo 1.0",
    "file": "z_image_turbo_1.0_q8p.ckpt",
    "prefix": "",
    "version": "zImage",
    "defaultScale": 16,
    "textEncoder": "qwen_3_vl_4b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "hiresFixScale": 24,
    "note": "[Z Image Turbo](https://huggingface.co/Tongyi-MAI/Z-Image-Turbo) is is a powerful and highly efficient image generation model with 6B parameters. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 8 sampling steps recommended.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Z Image Turbo 1.0 (6-bit)",
    "file": "z_image_turbo_1.0_q6p.ckpt",
    "prefix": "",
    "version": "zImage",
    "defaultScale": 16,
    "textEncoder": "qwen_3_vl_4b_instruct_q8p.ckpt",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "hiresFixScale": 24,
    "note": "[Z Image Turbo](https://huggingface.co/Tongyi-MAI/Z-Image-Turbo) is is a powerful and highly efficient image generation model with 6B parameters. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 8 sampling steps recommended.",
    "copyright": "© 2025 Alibaba"
  },
  {
    "name": "Z Image Turbo 1.0 (Exact)",
    "version": "z_image",
    "autoencoder": "flux_1_vae_f16.ckpt",
    "prefix": "",
    "default_scale": 16,
    "hires_fix_scale": 24,
    "file": "z_image_turbo_1.0_f16.ckpt",
    "upcast_attention": false,
    "text_encoder": "qwen_3_vl_4b_instruct_f16.ckpt",
    "high_precision_autoencoder": false,
    "objective": {
      "u": {
        "condition_scale": 1000
      }
    },
    "note": "[Z Image Turbo](https://huggingface.co/Tongyi-MAI/Z-Image-Turbo) is is a powerful and highly efficient image generation model with 6B parameters. It is Apache 2.0-licensed and commercially friendly. The model is trained at multiple resolutions using a Flow Matching objective; trailing samplers yield the best results, with 8 sampling steps recommended."
  }
]