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: fit reasoning_effort to a template's enforced allowlist (#16527)
5b219c9 Long Yixing committed 16 days ago
Studio: let managed accounts run terminal and Python tools on macOS (#16579)
ef65f8c Nilay committed 16 days ago
Tray status test: end the listener slice at its own close (#16600)
6f79197 Daniel Han committed 17 days ago
Parallel env tests: pin UNSLOTH_RETURN_LOGITS unset for the packing checks (#16594)
8d2d7b6 Daniel Han committed 17 days ago
Studio: faster MLX load and reload, ready to serve when loaded (#16536)
81dc73a Long Yixing committed 17 days ago
Bump install.sh / install.ps1 pins to unsloth>=2026.11.1 (#16554)
24d65c1 Daniel Han committed 17 days ago
Preserve prompt cache during deferred MCP tool search (#16546)
4710162 Wasim Yousef Said committed 17 days ago
Make page-in startup progress tests independent of host scheduling (#16476)
b773d91 Daniel Han committed 17 days ago
Studio diffusion: skip settling for proven resident replans (#16400)
eb648c2 Daniel Han committed 17 days ago
Keep Studio windows signed in during concurrent session refresh (#16468)
c1a6513 Daniel Han committed 17 days ago
Keep a verified live Desktop backend after watchdog recovery (#16470)
adfaf3c Daniel Han committed 17 days ago
Fix MoE adapter setup and reload with offloaded layers (#15159)
810c932 Leo Borcherding committed 17 days ago
Respect interface scaling in Settings search and rich-paste badge (#16471)
334d08d Daniel Han committed 17 days ago
Studio: keep the loaded model when Switch back has to download a GGUF (#16474)
7a462cc Nilay committed 17 days ago
Desktop: allow a custom backend port with UNSLOTH_DESKTOP_PORT (#16523)
a68651e Daniel Han committed 17 days ago
Studio: size audio.cpp TTS models from measured peaks and run TQ2_1 on the CPU (#16514)
28f3e5d oobabooga committed 17 days ago
Fix Connections dropdown shrinking on scroll and hover (#16543)
8482fc3 Wasim Yousef Said committed 17 days ago
Studio: combine OpenCode Console and Go into one connection (#16482)
d6ca61a Nilay committed 17 days ago
Multi-LoRA overflow test: read the kernel source as UTF-8 (#16544)
dadd36e Daniel Han committed 17 days ago
ROCm progress test: model a feed that resolves the no-torch pair (#16542)
9448b9d Daniel Han committed 17 days ago
MLX inference stub: answer the _FIRST_USE_WORK import zoo added (#16537)
867c424 Daniel Han committed 17 days ago
Prevent FP16 overflow in grouped multi-LoRA expert projections (#16414)
9a8b229 Daniel Han committed 17 days ago
Pipeline parallel: agree on overflow once when explicitly unscaling (#16417)
d0a16f9 Daniel Han committed 17 days ago
GPU supervisor signal test: publish its handshake files atomically (#16535)
6b3ea78 Daniel Han committed 17 days ago
FP8: keep rowwise _scaled_mm for float32 where cuBLAS supports it (torch < 2.14) (#16393)
3d99bf6 Daniel Han committed 17 days ago