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
Commits on main
Studio: no divider above the Live monitor's Detailed view row (#14972)
65c1ab9 Michael Han committed 29 days ago
Studio: give Intel XPU Triton a working C++ toolchain on Windows (#14814)
ff823ac Debojyoti Banerjee committed 29 days ago
Studio diffusion: report the split's primary card as the execution device for diffusers ModularPipeline (FLUX.3 on two cards) (#14917)
001c3c2 Daniel Han committed 29 days ago
CI: skip Windows, macOS and Tauri jobs on draft pull requests (#14971)
8a60386 Daniel Han committed 29 days ago
Studio: simpler API usage examples, and no Pi / Mistral Vibe outbound links (#14914)
4be98d1 Michael Han committed 29 days ago
Studio (AMD Linux): keep the bundled ROCm runtime when the system ROCm prepend hides the GPU (#14933)
1fd1640 Daniel Han committed 29 days ago
Installer (Windows): fall back to the python-build-standalone mirror for uv-managed Python (#14686)
5c1b054 Daniel Han committed 29 days ago
Studio setup: keep the extension-free mamba-ssm in the stale kernel check (#14687)
d33e77e Daniel Han committed 29 days ago
Studio: keep causal-conv1d for Qwen3.5 / Qwen3-Next remote implementations (#14685)
3e68722 Daniel Han committed 29 days ago
Studio (Windows): pair a kept torch 2.12 with triton-windows 3.7 (#14684)
8281152 Daniel Han committed 29 days ago
fix(export): pass trust_remote_code for checkpoint exports (#14722)
982eb96 Ryder committed 29 days ago
Studio: reuse Model Hub filter metadata between searches (#14265)
9ec8618 alkinun committed 29 days ago
Format dataset_media.py, merged without the repo formatter in #14840 (#14963)
61b2ffe Daniel Han committed 29 days ago
Check the weight index of dot-named Sentence Transformers module folders (#14952)
f172aec Daniel Han committed 29 days ago
Tests: follow the radius scale in the titlebar and remote access contracts (#14957)
5a04ef7 Daniel Han committed 29 days ago
Studio: pin managed dataset media URLs to the Hub and scope extracted archive members (#14840)
f443cf9 Daniel Han committed 29 days ago
Attention: keep packed fp32 on one xformers call, read the fp32 downcast from Q's device (#14805)
0947723 Daniel Han committed 29 days ago
DiffusionGemma: build the prefix cache per DataParallel replica (#14942)
02ddfff Daniel Han committed 29 days ago
Studio: raise the chat template limit from 64 KiB to 512 KiB (#14934)
85ee7d9 Daniel Han committed 29 days ago
Studio: preserve conversation archive order for turns with attachments (#14768)
bb89072 oobabooga committed 29 days ago
Studio: one fixed radius scale, applied to the shared primitives (#14919)
77949e6 Michael Han committed 29 days ago
Studio chat: go back to a chat's model when it is still loaded (#14568)
96121e6 Debojyoti Banerjee committed 29 days ago
FP8 gfx11 byte decoder: tighten comments from #14936 (#14947)
8d80091 Daniel Han committed 29 days ago