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: use browser Always allow rules during composer Full access (#13700)
c77695d Wasim Yousef Said committed 1 month ago
Chat UI test: close the stale page browser-side instead of waiting on its renderer (#13708)
23e7248 Daniel Han committed 1 month ago
CI: run lint checks in parallel and overlap independent setup in Windows smoke and Backend CI (#13710)
d975be3 Daniel Han committed 1 month ago
CI: overlap browser installs and cache restores with Install Unsloth in the UI smoke jobs (#13707)
ba87f8b Daniel Han committed 1 month ago
CI: cancel a pull request's leftover queued and running runs when it closes (#13709)
9a4ac1c Daniel Han committed 1 month ago
Skip the per-layer RoPE stream sync unless the model spans several devices (#13609)
1c0b34e Daniel Han committed 1 month ago
Skip xformers FA3 on H100 when the build has no sm_90a kernels (#13619)
11ee567 Michael Han committed 1 month ago
Fix Voxtral processor saving and Voxtral Realtime training (#12647)
453499b Yashwanth Nomula committed 1 month ago
Keep the pad embedding row when the tokenizer emits the pad token, never zero an EOS row (#13628)
4ec4b4c Daniel Han committed 1 month ago
Fix DPO on models split across GPUs with TRL 1.15's fused LM head (#13641)
3370575 Daniel Han committed 1 month ago
Unsloth Studio: redact the credential after an Authorization scheme in research queries (#13499)
504be6f Leo Borcherding committed 1 month ago
CI: bound CodeQL overlay-base cache churn, unstarve the cache janitor (#13704)
ad83778 Daniel Han committed 1 month ago
CI: coalesce main-push runs per workflow so merge bursts stop outrunning the runner pool (#13705)
80157ca Daniel Han committed 1 month ago
Fix main CI: rewind and Skills strings, token count harness, deep research contract, zoo 2026.10.5 scan baseline (#13701)
eb49b6f Daniel Han committed 1 month ago
Pack Qwen3.5 text SFT with the processor, and fix eval_loss on packed text runs (#13624)
3822d70 Daniel Han committed 1 month ago
Studio: Settings dropdowns glow in the pane colour in dark mode (#13703)
7f24926 Michael Han committed 1 month ago
Studio: let JSON-reply API requests still call their tools on MLX models (#13658)
42625f0 Nilay committed 1 month ago
Studio: reorder the chat + menu, with Deep research and Browser use (#13679)
34b84f7 Michael Han committed 1 month ago
Studio: keep SSE MCP servers as SSE when importing a data recipe (#13659)
2a03c22 Nilay committed 1 month ago
Studio: let read aloud go past 2048 tokens and warn when cut short (#13648)
b44035a Nilay committed 1 month ago
Studio: chat with files citation chips open a PDF at the cited page (#13691)
811eaa0 Michael Han committed 1 month ago
Studio: text #0f1114 in light mode and #fcfcfc in dark, thinking text easier to read (#13693)
8eb8835 Michael Han committed 1 month ago
Studio: removing a composer attachment no longer reloads the others (#13677)
4e63ad8 Michael Han committed 1 month ago
Studio: simpler Rewind confirm, skipped for conversation-only rewinds (#13676)
2a2ca69 Michael Han committed 1 month ago
Drop the divider above a tool's created-files row (#13692)
fda9f1c Michael Han committed 1 month ago
Studio: collapse Chat with files to a row of pills, with Show all (#13686)
7122742 Michael Han committed 1 month ago
Declared-polls probe: count only requests this Studio serves (#13690)
a530055 Daniel Han committed 1 month ago
Keep torch's bundled flash on Blackwell for short documents under UNSLOTH_FLASH_ATTENTION_IMPL=auto (#13631)
95526e4 Daniel Han committed 1 month ago