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: let the owner share a connection with every account (#14365)
0a930d4 Daniel Han committed 1 month ago
FP8 block dequant: ragged / rectangular e4m3fn blocks take the byte decoder on every NVIDIA GPU (4-16x) (#14450)
1c299c5 Daniel Han committed 1 month ago
Studio: picking a preset no longer overwrites a model's remembered sampling (#14128)
c1aef2f Daniel Han committed 1 month ago
Studio: keep lone asterisks and backticks in finished chat replies (#14391)
acb31d9 Nilay committed 1 month ago
Studio: trim the runtime update card to one summary line and two actions (#14436)
2af56e1 oobabooga committed 1 month ago
MoE grouped GEMM: accumulate expert dW into an existing param.grad (#14480)
d2f83d5 Daniel Han committed 1 month ago
Sample Qwen3 models for the turn's thinking mode in unsloth chat (#14388)
b368686 Nilay committed 1 month ago
Studio: stop two Studio windows undoing each other's browser data (#14277)
ba10ee8 Nilay committed 1 month ago
Studio: keep tools working after opening a chat file in its app (#14390)
4962983 Nilay committed 1 month ago
Studio: save files from download buttons inside browser panel pages (#14285)
8e95c16 Nilay committed 1 month ago
Studio: never walk init's descendants when stopping a server (#14486)
6446636 Daniel Han committed 1 month ago
Studio: honor completion-only training for audio VLMs (#14402)
5641b2a cherry77-cloud committed 1 month ago
Dependabot: leave hand-kept pins to the version matrix and hold the GTK3 and deep-link crates (#14470)
2b1d292 Daniel Han committed 1 month ago
Respect PIP_INDEX_URL/UV_INDEX_URL in update version check (#13748)
93fda20 SIDDARTHA REDDY committed 1 month ago
Studio: initialise the CUDA driver off the GIL so load-progress polls stay responsive (#14452)
fe8aab3 Daniel Han committed 1 month ago
Studio: keep the chat model loaded when an API request loads a speech model (#14369)
00cf280 Daniel Han committed 1 month ago
Desktop and Windows installer: browser-panel capture prompt, private ACL for a custom Studio home (#14464)
491d3dc Daniel Han committed 1 month ago
Core: per-user fp8 cache, SyntheticDataKit binds vLLM to loopback (#14460)
4be4f79 Daniel Han committed 1 month ago
Studio: keep Insert > Symbol characters when reading Word files (#14393)
b272568 Nilay committed 1 month ago
Studio: Z-Image unload test works on diffusers without ROPE_PER_DEVICE (#14478)
82f8240 Daniel Han committed 1 month ago
Studio: count reclaimable active file cache as available host RAM under a cgroup limit (#14475)
dcf5c40 Daniel Han committed 1 month ago
Studio: point Desktop Chat only users at Repair installation on the Export page (#14438)
93eed9e oobabooga committed 1 month ago
Studio Images: say why LoRAs vanish after Speed=Auto compiles, offer a reload (#14359)
d18a0a8 Daniel Han committed 1 month ago
Bump the cargo-tauri group in /studio/src-tauri with 26 updates (#14468)
e8f9004 dependabot[bot] committed 1 month ago
Studio: let browser pages open target=_blank links and show alerts (#14360)
1874390 Daniel Han committed 1 month ago
Unsloth Studio: redesigned API page, multi-model support and a Run settings fix (#14455)
97d38c7 Michael Han committed 1 month ago
Studio: one slider look everywhere, filled with the accent (#14471)
625522b Michael Han committed 1 month ago
Studio: managed accounts keep the configured Decision checkpoint; signed-out Hub relay hides upstream errors (#14462)
782fa2a Daniel Han committed 1 month ago
Studio: truncate over-length /v1/embeddings inputs instead of failing (#14364)
439411c Daniel Han committed 1 month ago