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
Improve model-load error guidance and shorten failure notifications (#16540)
8f2ba48 Lee Jackson committed 16 days ago
feat(studio): price GLM-5.3-Flash oQ quants as unsloth-zoo loads them (#16619)
6938f32 Long Yixing committed 16 days ago
Studio: check the weight index of a module path that only reads as the root (#16212)
c917be4 Daniel Han committed 16 days ago
Studio: keep native tool calls in mixed ShareGPT chats and render null-content calls on Qwen3 (#16120)
95383f9 Daniel Han committed 16 days ago
Block compile: checkpoint recompute reuses the forward graph (shared_kv_states, output collector) (#16596)
161ed47 Daniel Han committed 16 days ago
Join the Windows glass sidebar and title bar over chat wallpaper (#16641)
cec7ef0 Lee Jackson committed 16 days ago
fix(studio): load the MLX training dataset before the model (#16618)
d63d056 Long Yixing committed 16 days ago
Studio: remember download answers for a site or every site, programs included (#16658)
eec7711 Michael Han committed 16 days ago
Studio: let browser use read, click and edit files in the browser panel (#16644)
71c643e Michael Han committed 16 days ago
Studio: show Run settings gear on every GGUF quant row (#16660)
707851b Michael Han committed 16 days ago
Studio: glass theme polish and a composer position slider (#16643)
c2ec31f Michael Han committed 16 days ago
Diffusion tests: pin the NVIDIA-host spoof and assert ROCm keeps the stock path (#16609)
c15ac9b Daniel Han committed 16 days ago
Studio: size streamed SDXL UNet leaves from the denoise peak and lend them to the VAE decode (#16620)
38a8d12 Daniel Han committed 16 days ago
Studio: LiquidAI d1 and Decider decision models, train LFM3 into d1 (#16519)
e27c680 Long Yixing committed 16 days ago
Studio: report bad decision datasets before the model downloads (#16498)
096fa32 Nilay committed 16 days ago
MLX: run DPO, ORPO, KTO, GRPO and CPO through trl, and stop blocking SFT arguments zoo honors (#16461)
88fe11e Long Yixing committed 16 days ago
Studio: show the newest Kimi models in the model picker (#16591)
37e9429 Nilay committed 16 days ago
Studio diffusion: show whether group offload keeps the transformer on the GPU or streams it every step (#16610)
e36352f oobabooga committed 16 days ago
Studio: keep custom spellings when another window saves voice settings (#16577)
f7eeafb Nilay committed 16 days ago
Studio: keep non-English text in structured Laya instructions (#16510)
4d77df2 Nilay committed 16 days ago
Studio: let Laya fine-tunes answer from their GGUF export (#16511)
8bf2d3c Nilay committed 16 days ago
Studio: show status between tool rounds and while context compaction runs (#16608)
f2d72cb oobabooga committed 16 days ago
Studio: open Try after Use in Decision API for Clef and LLM fine-tunes (#16497)
55b6890 Nilay committed 16 days ago
Studio: ask Laya questions without instructions by their id (#16496)
d0a83c4 Nilay committed 16 days ago
Studio: list third-party image LoRAs from model folders (#16483)
4d85a6a Nilay committed 16 days ago