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: keep GGUF compaction when the catalog row goes stale (#10158)
7da85a0 Kenshi committed 2 months ago
Studio: say only localhost loses keyless access while a tunnel is up (#9690)
7b9bb03 Daniel Han committed 2 months ago
Stop the uninstaller stranding old installs, and deleting directories that are not ours (#10471)
1ad4467 Daniel Han committed 2 months ago
Bump install.sh / install.ps1 pins to unsloth>=2026.9.4, unsloth-zoo>=2026.9.3 (#10640)
191b69c Daniel Han committed 2 months ago
Studio: account for resident memory during model reloads (#10627)
ab28823 Michael Han committed 2 months ago
Studio: stop a quit during a model load from orphaning llama-server (#10592)
fcaad20 Daniel Han committed 2 months ago
Windows installer: shorten the native-type comments (#10621)
3cb17d7 Daniel Han committed 2 months ago
fix(studio): restore eager Settings and keep startup splash until ready (#10629)
e3e457a Wasim Yousef Said committed 2 months ago
Studio: restore the CLI help text the comment passes had trimmed (#10622)
3f2c895 Daniel Han committed 2 months ago
Windows: run the native-type tests on Windows, and fix the two that could not (#10624)
a73ecae Daniel Han committed 2 months ago
Studio: import readFileSync in the chat continuation test (#10628)
ec5c3ef Daniel Han committed 2 months ago
Restore what actually fails when NPP is missing from the torchcodec install (#10612)
0171568 Daniel Han committed 2 months ago
Studio: cover the unsigned 64-bit edge of a linked-folder identity (#10606)
6d6ca80 Daniel Han committed 2 months ago
Windows installer: emit the native path helper instead of compiling it (#10540)
95feb69 Daniel Han committed 2 months ago
Studio: try every address a site resolves to when fetching a page (#10319)
c12a37f Nilay committed 2 months ago
Studio: remove the background staged desktop update and return to the classic update (#10604)
f9acb70 Daniel Han committed 2 months ago
Studio: capture the diffusion denoiser step into a CUDA graph (#10537)
178e172 Daniel Han committed 2 months ago
Studio: give the conversation archive's order a final tiebreaker (#10572)
34120fb Daniel Han committed 2 months ago
[Fix] Studio: show "Using tool" for active tool calls (#10548)
29463b2 Debojyoti Banerjee committed 2 months ago
Share builders and setup helpers in the studio backend tests, part 1 (#10497)
4558ee7 Daniel Han committed 2 months ago
Studio: report a Claude reply cut off by the context window as truncated (#10557)
dc1b0be Nilay committed 2 months ago
Ask for the NPP runtime by the name that has wheels on CUDA 13 (#10587)
ad67389 Daniel Han committed 2 months ago
Guard main against drifting out of its own formatting hook (#10603)
19c86b3 Daniel Han committed 2 months ago
Studio: avoid hidden PowerShell arguments in CLI setup and updates (#10560)
247e3e8 oobabooga committed 2 months ago
Studio: skip torchcodec for torch below the compatibility table (#10506)
3779ac4 Rudra Bhaskar committed 2 months ago
Studio: find the AMD Vulkan driver through the Windows device registrations (#10564)
0f58608 oobabooga committed 2 months ago
Record the Windows GPU detection work from this branch (#5336)
61f01a3 Aron Rubin committed 2 months ago