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: synchronise VAE decode and encode calls on Apple MPS (#14940)
5e5ffb9 Long Yixing committed 29 days ago
Bump install.sh / install.ps1 pins to unsloth>=2026.10.8, unsloth-zoo>=2026.10.8 (#14994)
e9e7fe8 Daniel Han committed 29 days ago
Studio: cancel + bound conversation-archive embeddings before inference (#14519) (#14546)
7dbc8b9 Indrajit swain committed 29 days ago
fix(studio): use the target's embedded MTP head when the MLX checkpoint keeps it under the language model (#14974)
c5482a5 Long Yixing committed 29 days ago
Fix QLoRA on WavLM: keep its attention projections out of 4-bit (#14960)
9f86c49 Daniel Han committed 29 days ago
Studio: split Z-Image guided steps only while the denoiser is compiled or replaying graphs (#14868)
ae3dc79 Daniel Han committed 29 days ago
Merge branch 'main' of https://github.com/unslothai/unsloth
389be76 Daniel Han committed 29 days ago
Merge branch 'main' of https://github.com/unslothai/unsloth
bb6e59e Daniel Han committed 29 days ago
Windows: read safetensors shards with pread on transformers 5.0 - 5.16 (#14856)
6108ff5 Daniel Han committed 29 days ago
Studio: fix host memory checks and improve logging for GGUF image loads (#14514)
03639b7 Mathias Müller committed 29 days ago
Studio sd.cpp: keep sd.cpp's all-CPU thread count on small SMT hosts nothing restricts (#14847)
0fbe052 Daniel Han committed 29 days ago
Studio: refuse to start Claude Code or OpenCode on a context too small for their own prompt (#14799)
681bbb8 Nilay committed 29 days ago
Studio: keep turns an external provider's compaction drops searchable (#14731)
744ee65 Debojyoti Banerjee committed 29 days ago
Studio: let HTTP/SSE MCP tool calls run for the whole Max Tool Call Duration (#14308)
6950c1e Sunny B committed 29 days ago
Studio diffusion: record each card's CUDA graphs on its own capture stream, so compiled renders on a second GPU are correct (#14915)
abe4383 Daniel Han committed 29 days ago
Studio: dataset image guard keeps owner flows working (opt-in, WSL paths, unencoded spaces) (#14966)
fb104d2 Daniel Han committed 29 days ago
FP8 e4m3fn scale dequant: pass enable_fp_fusion like every other launch (#14969)
d954720 Daniel Han committed 29 days ago
Tests: stub _unmerge_flags_on_error in the merged hub destination harness (#14961)
6875840 Daniel Han committed 29 days ago
Studio: show when a chat is waiting for tool approval (#14877)
39f9ef6 oobabooga committed 29 days ago
Studio diffusion: CFG-parallel router keeps the wrapped forward's signature (FLUX.3 under CFG-parallel) (#14918)
b3004af Daniel Han committed 29 days ago
Unsloth Studio: browser use fixes and clearer agent progress (#14681)
55785e1 Michael Han committed 29 days ago
Studio: stop the macOS desktop app typing stray control characters for unhandled keys (#14815)
394aff8 Debojyoti Banerjee committed 29 days ago
Studio: attach duplicate-named text files like LICENSE(1), and say why a file can't be attached (#14977)
bf17205 Michael Han committed 29 days ago
Load and QLoRA-train Moonshine Streaming checkpoints (#14945)
a3e7ac7 Daniel Han committed 29 days ago
Studio: drop file toolbar controls that repeat a visible button (#14976)
ac49edd Michael Han committed 29 days ago
Studio: keep renamed audio.cpp options at generation time (#14901)
472b1ca Daniel Han committed 29 days ago