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
fix(studio): reject ambiguous S3 audio references (#16529)
d454907 Max-Reisinger committed 16 days ago
Studio: keep long replies with escaped math streaming smoothly (#16490)
b87e219 Nilay committed 16 days ago
Forward the model argument through update_weight_conversions overrides (transformers 5.21+) (#16559)
ca762a0 Daniel Han committed 16 days ago
Studio: attach saved Library files to the current chat (#16522)
220719e Daniel Han committed 16 days ago
Studio: keep image follow-ups running after the tab closes (#16491)
70f11ee Nilay committed 16 days ago
Backend CI: run tensor and pipeline parallel gloo suites in separate shards (#16634)
624cd34 Daniel Han committed 16 days ago
Studio: do not download PyTorch Clef for an over-long decision request (#16509)
a970757 Nilay committed 16 days ago
Studio: stop showing a cancelled tool request reply as finished (#16473)
67e82c8 Nilay committed 16 days ago
Studio: scale spreadsheet figures in thousands or millions with decimals (#16590)
f3892ae Nilay committed 16 days ago
Studio: keep profile and appearance changes made in another window (#16580)
f5adb36 Nilay committed 16 days ago
Multi-LoRA: export expert LoRA in the layout vanilla PEFT loads (#16568)
02c8fa4 Daniel Han committed 16 days ago
FP8 rowwise scaled_mm: keep fp16 output in range before torch 2.14, and drop the row guard for always-dequant dtypes when compiled (#16598)
1e4643c Daniel Han committed 16 days ago
Studio: let the desktop app stop a server it attached to, and show its port (#16595)
f10c0ca oobabooga committed 16 days ago
Fix duplicated wallpaper titlebar tint and modal dimming (#16571)
f6c13d4 Lee Jackson committed 16 days ago
Hide the sidebar footer fade only in glass themes (#16572)
73a283f Lee Jackson committed 16 days ago
Context parallelism: generic Ulysses on any Hugging Face attention model, per-instance trainer wrappers (#16558)
7dd74a5 Daniel Han committed 16 days ago
Scan baseline: re-review the five unsloth-zoo sites 2026.11.1 moved (#16624)
0463e4e Daniel Han committed 16 days ago
Studio diffusion kernels: keep per-resolution sizes out of the Triton compile key (#16582)
525e4ab Daniel Han committed 16 days ago
fix(studio): price MLX attention on the route the runtime takes (#16517)
4691cc6 Long Yixing committed 16 days ago
Studio: stop browser panel pages typing stray control keys on macOS (#16583)
42aeac4 Nilay committed 16 days ago
Kernels: keep per-call sizes out of the Triton compile key (layernorm dW, fp8 block matmul, nvfp4 split sum) (#16584)
c15435d Daniel Han committed 16 days ago
Kernels: name every tl.constexpr parameter with a trailing comment (#16585)
5a13ca9 Daniel Han committed 16 days ago