From c4f2fb1dfd3dcecc9ffe317791a86afaf0cd982f Mon Sep 17 00:00:00 2001 From: hekmon Date: Thu, 15 Oct 2026 13:16:48 +0000 Subject: [PATCH] docs: replace deprecated TGI with SGLang The TGI repository was archived on 2026-03-21 and is in maintenance mode since late 2025; Hugging Face now recommends vLLM and SGLang as forward inference engines. --- README.md | 2 +- pyproject.toml | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index a4ef2b4..b8c3f1e 100644 --- a/README.md +++ b/README.md @@ -12,7 +12,7 @@ Thanks to its simplicity the project has a low footprint: it only has 1 external ## Usage -The default `base_url` parameter value targets the official OpenAI API endpoint by default but by changing it, you can also use this project with any OpenAI API compatible servers like Ollama, vLLM, TGI, etc... +The default `base_url` parameter value targets the official OpenAI API endpoint by default but by changing it, you can also use this project with any OpenAI API compatible servers like Ollama, vLLM, SGLang, etc... Multiples images are supported as long as they are fed batched to the chat completion node. diff --git a/pyproject.toml b/pyproject.toml index 042f8b7..f0d412b 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "openai-api" -description = "Call LLM and VLM from ComfyUI using the OpenAI API standard. Compatible with custom inference servers like Ollama, vLLM or TGI thru their support of their OpenAI API implementation." +description = "Call LLM and VLM from ComfyUI using the OpenAI API standard. Compatible with custom inference servers like Ollama, vLLM or SGLang thru their support of the OpenAI API implementation." version = "3.0.0" license = { file = "LICENSE" } # Numpy and Pillow same as ComfyUI