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 diffusion: off-GIL CUDA prime does not latch before torch is imported (#14683)
0c2dfda Daniel Han committed 24 days ago
Studio diffusion: CUDA graph re-check fixes (eager OOM retry on originals, planned negative length records on the first render) (#14682)
c519609 Daniel Han committed 24 days ago
Keep RAG WAL sidecars between database operations (#15719)
47f064c Lee Jackson committed 24 days ago
Studio: drop three frontend dependencies nothing imports (#15634)
9b315cb Daniel Han committed 24 days ago
Studio desktop: build release with fat LTO and one codegen unit (#15566)
2a2f31c Daniel Han committed 24 days ago
Studio: Search and Tools on by default for local models (#15644)
43bc7c4 Michael Han committed 24 days ago
Studio: remove private helpers and constants that nothing uses (#15696)
1f55d81 Daniel Han committed 24 days ago
Studio: start the torch warm-up stages after the first response is sent (#15732)
6b9b6c2 Daniel Han committed 24 days ago
Tests: follow the media picker's Chat-aligned height, start pwsh through run_pwsh (#15758)
947358e Daniel Han committed 24 days ago
Studio desktop: hide the window as soon as a quit is committed, and exit the backend without finalization (#15723)
a7a1a85 Long Yixing committed 24 days ago
Saving: stop repeated gc.collect passes once a pass finds nothing (#15694)
3d7e798 Daniel Han committed 24 days ago
Update OpenEnv links to openenv/OpenEnv (#15743)
6ccc84f Sergio Paniego Blanco committed 24 days ago
Add AGPLv3 header to desktop release workflow (#15755)
e8150f6 Wasim Yousef Said committed 24 days ago
CLI: load command modules other than studio only when a command is dispatched (#15689)
83f76e2 Daniel Han committed 24 days ago
Studio: fewer GC passes while the backend imports, RSA key generated alongside startup (#15586)
7362fbc Daniel Han committed 24 days ago
Packed attention: full finetuning rows reach varlen (fp32 Q / K under autocast) (#15714)
afa0234 Daniel Han committed 24 days ago
Fix Gallery save toast to use the chosen filename (#15725)
fe032ba Lee Jackson committed 24 days ago
Studio: read the VLM and audio detection sets from the transformers registry source instead of importing torch (#15688)
94be55a Daniel Han committed 24 days ago
Studio: Trick or treat sidebar row with Halloween themes and spooky greetings (#15678)
39b8a6c Michael Han committed 24 days ago
Studio: keep a Windows console stderr writable while the worker mirror owns fd 2 (#15734)
eb53671 Daniel Han committed 24 days ago
Studio: web fetch tokenizes a page once and replays it to the later main-content passes (#15686)
bebcc23 Daniel Han committed 24 days ago
Studio: refresh cache sizes that are being re-measured instead of showing the old number (#15718)
ed9857c Daniel Han committed 24 days ago
Studio: reuse VAE tile blend weights in tiled encode and LTX-2 tiled decode (#15700)
87ea22f Daniel Han committed 24 days ago
Studio: let the router skip routes whose path prefix cannot match (#15682)
bcf0365 Daniel Han committed 24 days ago
Dependencies: drop four Python packages nothing imports (tyro, flatbuffers, pytorch_tokenizers, mammoth) (#15697)
b7a110f Daniel Han committed 24 days ago
setup.sh: keep the replaced-script handoff directly behind the dependency pass (#15747)
4b71ee3 Daniel Han committed 24 days ago
Studio: bound the passthrough healer gate to openers the parser would scan (#15717)
41c6a0d Daniel Han committed 24 days ago
Studio training: log the traceback when a model fails to load (#15620)
af068cc Debojyoti Banerjee committed 24 days ago