Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
Python agentaichatgptdeepseekfine-tuninggemmaimage-generationllamallmllmsopenaipythonqwenreinforcement-learningself-hostedstable-diffusiontext-to-speechttsuiunsloth
Unsloth Studio: raise the huggingface_hub floor to what the pinned Diffusers main build needs (#11625)
Leo Borcherding78fbee7597a2
diff --git a/pyproject.toml b/pyproject.tomlindex 1d65be2df0b..075a2366cdd 100644--- a/pyproject.toml+++ b/pyproject.toml@@ -155,7 +155,7 @@ studio = ["nest_asyncio==1.6.0","datasets==4.3.0","pyjwt==2.13.0",- "huggingface-hub>=1.23.0,<2.0 ; python_version >= '3.10'",+ "huggingface-hub>=1.31.0,<2.0 ; python_version >= '3.10'","huggingface-hub==0.36.2 ; python_version < '3.10'","hf-xet>=1.5.2,<2.0 ; python_version >= '3.10' and (platform_machine == 'x86_64' or platform_machine == 'amd64' or platform_machine == 'AMD64' or platform_machine == 'arm64' or platform_machine == 'aarch64' or platform_machine == 'ARM64')","urllib3>=2.3.0",diff --git a/studio/backend/requirements/no-torch-runtime.txt b/studio/backend/requirements/no-torch-runtime.txtindex 5b1716b6a33..42f1dcb561a 100644--- a/studio/backend/requirements/no-torch-runtime.txt+++ b/studio/backend/requirements/no-torch-runtime.txt@@ -46,7 +46,7 @@ safetensors>=0.4.3datasets>=3.4.1,!=4.0.*,!=4.1.0,<4.4.0accelerate>=0.34.1peft>=0.18.0,!=0.11.0-huggingface_hub>=1.23.0,<2.0; python_version >= "3.10"+huggingface_hub>=1.31.0,<2.0; python_version >= "3.10"huggingface_hub==0.36.2; python_version < "3.10"# --no-deps here, so the hub never pulls it transitively, and without it huggingface_hub refuses any# single blob over 50GB (file_download.py, MAX_HTTP_DOWNLOAD_SIZE). Arch gate is huggingface-hub'sdiff --git a/studio/backend/requirements/single-env/constraints.txt b/studio/backend/requirements/single-env/constraints.txtindex f0d2cbd2358..0de0d05442e 100644--- a/studio/backend/requirements/single-env/constraints.txt+++ b/studio/backend/requirements/single-env/constraints.txt@@ -8,7 +8,7 @@ transformers==4.57.6; python_version < "3.10"# transformers pin, from pairing them with a tokenizers transformers rejects.tokenizers>=0.22.0,<=0.23.0trl==0.23.1-huggingface-hub>=1.23.0,<2.0; python_version >= "3.10"+huggingface-hub>=1.31.0,<2.0; python_version >= "3.10"huggingface-hub==0.36.2; python_version < "3.10"hf-xet>=1.5.2,<2.0; python_version >= "3.10" and (platform_machine == "x86_64" or platform_machine == "amd64" or platform_machine == "AMD64" or platform_machine == "arm64" or platform_machine == "aarch64" or platform_machine == "ARM64")diff --git a/studio/backend/requirements/single-env/overrides-darwin-arm64.txt b/studio/backend/requirements/single-env/overrides-darwin-arm64.txtindex 90e5ce8d608..897bd1f8c42 100644--- a/studio/backend/requirements/single-env/overrides-darwin-arm64.txt+++ b/studio/backend/requirements/single-env/overrides-darwin-arm64.txt@@ -35,13 +35,9 @@ anyio<4.14.0# silently backtracks mlx-vlm to 0.4.4, dropping newer VLM/audio support, so the# cap stays: the resolver must never pair hub 1.x with pinned transformers 4.x.#-# On python >= 3.10 that premise is gone. diffusers (for MiniMax-H3) requires-# huggingface-hub>=1.23.0,<2.0, and constraints.txt, studio.txt, base.txt and-# no-torch-runtime.txt all moved to that window. A flat <1.0 here contradicts-# them and is unsatisfiable, which is macOS-arm only, so only the mac legs see-# it: uv reports "Because you require huggingface-hub>=0.34.0,<1.0 and-# huggingface-hub>=1.23.0,<2.0, your requirements are unsatisfiable", falls back-# to pip, and the clean-machine trace then fails on a rustc probe. mlx-audio's-# own >=1.0 floor is satisfied by the 1.x window anyway.-huggingface_hub>=1.23.0,<2.0; python_version >= "3.10"+# On python >= 3.10 the premise is gone: constraints.txt, studio.txt,+# no-torch-runtime.txt and the pyproject studio extra take hub 1.x, so a flat+# <1.0 here is unsatisfiable (macOS-arm only; uv falls back to pip). 1.31.0 is+# the floor the Diffusers main build in diffusers-main.txt declares.+huggingface_hub>=1.31.0,<2.0; python_version >= "3.10"huggingface_hub>=0.34.0,<1.0; python_version < "3.10"diff --git a/studio/backend/requirements/studio.txt b/studio/backend/requirements/studio.txtindex 6a1c8f2af62..adff3c13320 100644--- a/studio/backend/requirements/studio.txt+++ b/studio/backend/requirements/studio.txt@@ -20,7 +20,7 @@ nest_asyncio==1.6.0datasets==4.3.0pyjwt==2.13.0# gradio>=4.0.0 # 148 MB - Unsloth uses React + FastAPI, not Gradio-huggingface-hub>=1.23.0,<2.0; python_version >= "3.10"+huggingface-hub>=1.31.0,<2.0; python_version >= "3.10"huggingface-hub==0.36.2; python_version < "3.10"# Matches no-torch-runtime.txt, the file that actually needs it; see there for the arch gate.hf-xet>=1.5.2,<2.0; python_version >= "3.10" and (platform_machine == "x86_64" or platform_machine == "amd64" or platform_machine == "AMD64" or platform_machine == "arm64" or platform_machine == "aarch64" or platform_machine == "ARM64")