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: load the speech model an audio API request asks for (#10217)
2e6db0c Nilay committed 2 months ago
Keep the LoRA GGUF and compressed-tensors converters off the operator's Hugging Face token (#10263)
5ae462d Sourav Rajvi committed 2 months ago
Studio: share one Hub model_info read across the model-config probes (#10230)
ff13134 Long Yixing committed 2 months ago
Studio: reuse MLX vision-language prompt prefixes between turns (#10239)
975c615 Long Yixing committed 2 months ago
Studio: give a local GGUF one variant identity, so an API load applies its saved settings (#10238)
c55dd94 Long Yixing committed 2 months ago
Studio: route AMD integrated GPUs to the Vulkan llama.cpp prebuilt (#10381)
8b36928 Daniel Han committed 2 months ago
Studio: do not crash a model load when shutdown tears the server down under it (#10369)
03dabb2 Daniel Han committed 2 months ago
Studio: optimize video galleries by removing per-card media pipelines (#10228)
f070971 oobabooga committed 2 months ago
Stop shipping the frontend source tree and desktop crate in the wheel (#10426)
fe1d83c Daniel Han committed 2 months ago
Studio: report a throughput the engine could have produced (#10384)
4679209 Daniel Han committed 2 months ago
Publish Docker on the daily schedule, and move layer cache off the Actions quota (#10429)
583191a Daniel Han committed 2 months ago
Studio: retry as one sequence when llama.cpp refuses a unified KV cache (#10371)
e37730d Daniel Han committed 2 months ago
Studio: keep the llama-server prompt cache on a shared-memory GPU (#10382)
c2bb6e3 Daniel Han committed 2 months ago
Studio: four load and update messages that told the user the wrong thing (#10383)
3a46167 Daniel Han committed 2 months ago
Studio installer: torch 2.11 on Linux CPU, Vulkan for AMD without ROCm, gfx1033 gate, and three install fixes (#8412)
e129630 Daniel Han committed 2 months ago
Give `unsloth studio update` the same uv cache the backend already uses (#10386)
b66d2a4 Daniel Han committed 2 months ago
Studio smoke: report a transport failure instead of crashing on one (#10420)
fb9bb4e Daniel Han committed 2 months ago
Studio: unbreak the frontend suite and the icon alignment check (#10413)
5b283e5 Daniel Han committed 2 months ago
Studio: make the chat and sidebar menus open without freezing the page (#10262)
b48b602 Nilay committed 2 months ago
Kaggle CI: dispatch the kernel and collect it later, instead of holding a runner (#10183)
6bc095a Daniel Han committed 2 months ago
Give push_to_ollama the arguments create_ollama_modelfile takes (#10304)
2a95eb7 Vineeth Sai Varikuntla committed 2 months ago
Fix composer send and stop icon centering across display scales (#10407)
a9aa135 Michael Han committed 2 months ago
Docker toolkit installer: stop pipefail turning a grep match into a miss (#10402)
a6e9880 Daniel Han committed 2 months ago
Local Agent Guides CI: judge a finished turn the CLI would not exit from (#10399)
dcf1237 Daniel Han committed 2 months ago
Studio: return an error when tools are sent to a model that cannot use them (#10314)
066290d Nilay committed 2 months ago
Studio: stop converting the model twice when exporting GGUF to the Hub (#10317)
7b1d8ed Nilay committed 2 months ago
Studio: serve /v1/embeddings from the configured embedding model when the loaded GGUF cannot (#10315)
b8d120a Nilay committed 2 months ago
Point unsloth start openclaw memory search at the Unsloth embeddings API (#10320)
083113b Nilay committed 2 months ago
feat(chat): render on-disk sandbox images in chat markdown (#10361)
df0c5ce CelesteHeartsong committed 2 months ago
Use hybrid uv cache migration for Studio installs (#10204)
919922c Wasim Yousef Said committed 2 months ago