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 Windows XPU Triton: verify the swap and torch's XPU triton payload, repair a broken triton, skip pip backups (#16250)
c72742f Daniel Han committed 18 days ago
Studio: wait on worker exit instead of fixed sleeps when switching models (#16344)
82e9740 Daniel Han committed 18 days ago
Studio: keep an anonymous Hub token anonymous in the hosted bf16 VAE path (#16214)
65f72db Daniel Han committed 18 days ago
Studio: do not subtract live segment graph outputs twice from reusable cache (#16231)
babbdbf Daniel Han committed 18 days ago
Studio: show the voice description on Clone for Breeze TTS 2 and VoxCPM2 (#15878)
ee30489 Etherl committed 18 days ago
Studio video: reuse bounded conditioning image decodes (#16403)
b58ea23 Daniel Han committed 18 days ago
Block compile: say when packed rows run each block uncompiled (#16272)
3d9b1cd Daniel Han committed 18 days ago
Fused LM head: accept TRL 1.17's score_ids kernel argument (#16220)
3ff5560 Daniel Han committed 18 days ago
Concretize RMSNorm width before the narrow-row self-check (#16228)
7cc7f98 Daniel Han committed 18 days ago
Studio: keep a nested active HF hub clearable and listed (#16215)
781f004 Daniel Han committed 18 days ago
Studio: keep feature indexes and the markdown renderer out of the startup bundle (#16394)
b36f53d Daniel Han committed 18 days ago
Studio: train audio VLMs on the images in a mixed audio + image dataset (#16124)
50f09c8 Leo Borcherding committed 18 days ago
Studio: apply the mapped chat template once per loaded tokenizer (#16377)
2f1cc71 Daniel Han committed 18 days ago
Studio: restore public OpenAPI descriptions on four routes and schemas (#16223)
24ce872 Daniel Han committed 18 days ago
Skip pinned staging for native int8 host quantisation on integrated GPUs (#16235)
9fd8533 Daniel Han committed 18 days ago
Fix reasoning classification for generation-tag chat templates (#16121)
8f5b576 Leo Borcherding committed 18 days ago
Studio: match diffusion training access facts to the no-token mirror fetch (#16216)
7af1d00 Daniel Han committed 18 days ago
Studio: keep a text selection in Settings › Logs across polls so part of the log can be copied (#16099)
35385ed Leo Borcherding committed 18 days ago
Studio: keep the chat background, interface scale and Browser preferences with the account (#16454)
bae35f6 Michael Han committed 18 days ago
Studio: paste links, tables, lists and formatting into the composer as Markdown (#16298)
63a3896 Michael Han committed 18 days ago
Recognize shared encoder assets in Model Hub inventory (#16466)
c57f7ea Lee Jackson committed 18 days ago
Fix pipeline denoiser downloads labeled as required assets (#16465)
40bb5d6 Lee Jackson committed 18 days ago
Studio: report the measured VRAM shortfall instead of blaming the previous worker (#16453)
a4becf8 Yakshith committed 18 days ago
studiobench parity: treat every data-*-id attribute as an instance id (#16464)
4f4305e Daniel Han committed 18 days ago
Honor Voice Settings device preference for STT API loads (#15284)
2099086 Indrajit swain committed 18 days ago
fix(api): preserve refusal text in collected connection replies (#16435)
98f96f0 Henry Su committed 18 days ago
Studio: translate the live generation figures, the reply timing tooltip and the weight placement row (#16452)
dcfc392 oobabooga committed 18 days ago
Studio: Composer appearance section, composer outline on by default (#16455)
adfd9e0 Michael Han committed 18 days ago