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
Docker ROCm Studio image: run GGUF chat on the GPU through Vulkan (#13852)
d65cc7b Daniel Han committed 1 month ago
Studio Hub: Recommended lists safetensors repos only past 30 likes, always open on Discover (#13886)
186a9bf Michael Han committed 1 month ago
Studio: stream /v1/responses from MLX and other in-process models (#13794)
4978410 Long Yixing committed 1 month ago
Format tests/test_triton_fp_fusion.py back to the hook's fixed point (#13887)
23f5825 Daniel Han committed 1 month ago
Compile each transformer block as one region for every FastModel model (LoRA and QLoRA) (#10744)
c69969d Hexu Zhao committed 1 month ago
Studio: info stamp on image downloads, recipe in every format, load settings from an image (#13444)
2470977 Daniel Han committed 1 month ago
fast_inference: switch Gemma 4 to Unsloth inference on GPUs below sm80 (#13864)
f6c34dd Daniel Han committed 1 month ago
Studio: show a download start the moment it is requested, and serialise the first zoo import with the media import window (#13738)
5337c8f Eyera committed 1 month ago
Fine-tune and export native MTP heads and DSpark drafters (#13719)
65d3ca5 Daniel Han committed 1 month ago
Add UnlimitedOcrTextModel to the fused-MoE conversion snapshot (#13867)
cc483b3 Daniel Han committed 1 month ago
Studio: install causal-conv1d and mamba-ssm only when a model needs them (#13861)
90ac3b7 Daniel Han committed 1 month ago
Desktop: explain Windows Long Paths before its UAC prompt (#13523)
9f5b133 Daniel Han committed 1 month ago
Kernels: request enable_fp_fusion=True explicitly, ahead of Triton 3.9's False default (#13828)
e1cdbf4 Daniel Han committed 1 month ago
Prebuilt CUDA wheels: compress the fatbins (flash-attn 1.49 GB -> ~285 MB) (#13854)
7c235de Daniel Han committed 1 month ago
Studio Desktop: open the OS folder dialog to add a model folder (#13843)
62e0636 Daniel Han committed 1 month ago
fast_inference: Gemma 4 Unified, Muse Glimmer, dense Gemma 4; 4-bit fallback when no -bnb-4bit repo (#13825)
dda71fd Daniel Han committed 1 month ago
Add dtype option to save_pretrained_merged for float32 merges (#13823)
f5f3c43 Daniel Han committed 1 month ago
Fix SDAR empty-mask resampling to match the reference (#13771)
1389ec4 cherry77-cloud committed 1 month ago
Load the zoo Hub download helpers without unsloth_zoo's package init (#13849)
f9d41b1 Daniel Han committed 1 month ago
Pass layers_pattern from FastLanguageModel.get_peft_model to LoraConfig (#13751)
e59b52e Vineeth Sai Varikuntla committed 1 month ago
Studio: let Orpheus TTS pick one of its built-in voices (#13853)
848d723 Daniel Han committed 1 month ago
Studio: accept sk-unsloth API keys in the x-api-key header (#13842)
6159f7e Daniel Han committed 1 month ago
Fix DiffusionGemma sampled-prefix cache retention (#13780)
3f502e2 cherry77-cloud committed 1 month ago
Studio: send audio.cpp request fields in the form each model reads (#13791)
402a405 Etherl committed 1 month ago
Run Linux frontend browser checks in two parallel groups (#13813)
fd7d7b0 Nathan Tran committed 1 month ago
unsloth run: add --engine-parallelism and --gpus for multi-GPU serving (#13838)
16ec451 Daniel Han committed 1 month ago
Studio recipes: keep a finished run's pump off the next run (#13845)
2a7c55f Daniel Han committed 1 month ago