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: vector cache in float32 unit rows with streamed builds, in-place appends and bounded block scans (#16093)
6fe27d2 Daniel Han committed 22 days ago
Studio: download the commit the manifest was planned from (#16047)
5344638 Daniel Han committed 22 days ago
Studio diffusion: Qwen-Image text-to-image streamed placement uses its own activation figure under a VRAM budget (#16095)
08b3239 Daniel Han committed 22 days ago
Studio diffusion: test-launch sd.cpp binaries without the Windows missing-DLL box (#16031)
982ab34 oobabooga committed 22 days ago
Studio: read streamed dataset slices in batches instead of row by row (#16083)
6a87eb7 Daniel Han committed 22 days ago
Studio Images: enable GGUF LoRAs on MLX and diffusers, and explain when unavailable (#16053)
6d5a4d8 oobabooga committed 22 days ago
Studio: share one in-flight media model index build so the startup checks do not scan twice (#16075)
baa84e6 Daniel Han committed 22 days ago
Studio installer (Windows): avoid unavailable CUDA torchao wheels (#16016)
8b974d6 oobabooga committed 22 days ago
Studio: default Deep Research to no time limit per model request (#15983)
d769846 Nilay committed 22 days ago
Studio: ask llama-server again with its tool parser off when it rejects a reply before streaming (#16002)
848676c Daniel Han committed 22 days ago
Studio: recover Deep Research from provider JSON validation failures (#15788)
e09758d alkinun committed 22 days ago
Desktop: switch the theme without a view transition on Linux (#16040)
87d1add oobabooga committed 22 days ago
Studio: keep attached documents in Deep Research reports (#16019)
7352b62 Nilay committed 22 days ago
Tests: follow the localized update banners from #16069 (#16090)
754bd21 Daniel Han committed 22 days ago
Studio: offer video input only when llama-server can launch ffmpeg and ffprobe (#16094)
4e799e0 Daniel Han committed 22 days ago
Studio: keep a resident GGUF's repo identity for Hub shared-blob cache files (#16086)
858a97c Daniel Han committed 22 days ago
Bump the npm-oxc-validator group in /studio/backend/core/data_recipe/oxc-validator with 3 updates (#15962)
ece29f8 dependabot[bot] committed 22 days ago
Bump github/codeql-action init and analyze to v4.39.1 (#15964)
545f580 dependabot[bot] committed 22 days ago
fix(studio): keep a long MLX vision chat's prompt snapshot past the cache budget (#16079)
80a660a Long Yixing committed 22 days ago
Studio: size a cached DFlash sidecar from disk instead of asking the Hub on every load (#15986)
c3c04e2 Daniel Han committed 22 days ago
Studio: int8 Qwen-Image streamed modulations through a fused GEMV (1.23x under a 16 GiB budget) (#16087)
fca19f5 Daniel Han committed 22 days ago
Studio: let managed accounts use local MCP servers when allowed (#16044)
082ad75 Nilay committed 22 days ago
Report measured model loading and 4-bit conversion during image training (#15536)
b48b16e Lee Jackson committed 22 days ago
Tests: run the MXFP4 operand device guard check on CPU-only torch too (#16088)
583e552 Daniel Han committed 22 days ago
Studio: read a JSON body inside <function=...> instead of running the call with no arguments (#16008)
d4f1c2a Daniel Han committed 22 days ago