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: head a decision score with its most likely level (#13111)
8e3fd25 Leo Borcherding committed 1 month ago
Fix returned logit scaling and softcapping when labels are provided (#13069)
9d0f0e1 cherry77-cloud committed 1 month ago
Studio: show the context meter for active project chats (#12605)
45c7524 lorenzozane committed 1 month ago
fix(chat): import branch threads parents-before-children on refresh (#13113)
bc3c978 CelesteHeartsong committed 1 month ago
Studio: let Gemma 4 read web search and tool results on safetensors (#13096)
20e12f5 Nilay committed 1 month ago
Studio: let Deep Research search MCP tools as extra sources (#13103)
eca9573 Nilay committed 1 month ago
Studio: open Images workflows from More when Images is unpinned (#13101)
8016226 Nilay committed 1 month ago
Studio: read news articles whose pages carry megabytes of inline code (#13100)
83c7cef Nilay committed 1 month ago
Studio: hold message hover still while a wheel scroll moves the thread (#13077)
064a873 Daniel Han committed 1 month ago
Studio: let imported chats with image links answer follow-up questions (#13099)
c2c197d Nilay committed 1 month ago
Studio: raise the llama-server micro-batch to 2048 when MoE experts spill to RAM (#12950)
3ac616d Daniel Han committed 1 month ago
Studio: keep Think off working on Claude Opus 5.5 and Sonnet 5.5 (#13104)
264eec5 Nilay committed 1 month ago
Studio: stop link URLs from crowding out long page text (#13102)
ac6be24 Nilay committed 1 month ago
Studio: keep table columns in attached Word and HTML files (#13105)
8209ac4 Nilay committed 1 month ago
Windows installer: stop installing Git and the VS toolchain without asking (#12971)
c1302ee Leo Borcherding committed 1 month ago
GGUF adapter export and llama.cpp installer: use unslothai/llama.cpp releases only, no git (#13056)
db40043 Daniel Han committed 1 month ago
Studio: let Gemma 4 GGUF models think after a tool call (#13097)
9e5d03b Nilay committed 1 month ago
Studio: provide Git lazily for FP8/FP4 export instead of at install time (#13084)
6b833e3 Daniel Han committed 1 month ago
Reject Activated LoRA adapters instead of training them as plain LoRA (#13089)
5333c96 Daniel Han committed 1 month ago
Studio: keep typing and streaming fast in long chats (#13059)
63988fd Daniel Han committed 1 month ago
Unsloth Studio / Desktop: per-GPU Block Swap budget and layer panel for multi-GPU training (#12998)
85ae421 Leo Borcherding committed 1 month ago
Unsloth Studio / Desktop: Block Swap settings, a VRAM budget, and a live layer panel on the training page (#12996)
d953dca Leo Borcherding committed 1 month ago
Dequantize pre-quantized bitsandbytes checkpoints for full finetuning (#13079)
7a11957 Daniel Han committed 1 month ago
fix(studio): pad VAE downsample shortcuts by concatenation on MPS (#13028)
92cbd1d Long Yixing committed 1 month ago
Studio: load community SDXL single-file GGUFs on the Images page (#13035)
f9718db Daniel Han committed 1 month ago