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 video gallery: clear() reads and unlinks sidecars in the folder it locked (#15766)
a627f97 Daniel Han committed 24 days ago
Studio: skip the DFlash size request when no sidecar is eligible (#15754)
d4e8ae8 Daniel Han committed 24 days ago
Studio: hash the tokenizer fingerprint without touching the live tokenizer (#15749)
b4e5559 Daniel Han committed 24 days ago
Studio: decode each relayed SSE line once in the tool call stripper (#15753)
541e5a8 Daniel Han committed 24 days ago
Studio: load the attachment hasher before deleting a chat attachment (#15751)
90a6543 Daniel Han committed 24 days ago
Studio: read the Int8Tensor constructor signature once per class in the native INT8 rebuild (#15701)
6599e4a Daniel Han committed 24 days ago
Installer downloads: retry a body cut short or trickling instead of hashing it (#15763)
93426bf Daniel Han committed 24 days ago
Keep packed documents apart under flash attention when layers get mRoPE position ids (#15771)
3b01845 Daniel Han committed 24 days ago
Studio: batch packed fine-tuning jobs, each job packing only its own documents (#15733)
75555cc Daniel Han committed 24 days ago
Clarify Deep Research handoff for research briefs (#15724)
41443a4 Lee Jackson committed 24 days ago
Studio: preview JSON Lines datasets with the stdlib, not pyarrow's JSON reader (#15713)
998661d Shubhankar Tripathy committed 24 days ago
Studio: cast-cache writer reads the host copy offload already keeps instead of waiting out an onload (#15698)
d4e56bc Daniel Han committed 24 days ago
Studio: spider or bats on a third of Halloween pages each (#15770)
7065bf6 Michael Han committed 24 days ago
Studio: speedometer icon for generation speed on the API page (#15769)
52ced46 Michael Han committed 24 days ago
cuDNN SDPA grad layout fix: let compiled code modify the SDPA output in place (#15721)
0512f4e Daniel Han committed 24 days ago
FSDP2 block CUDA graphs: free the gathered weights of MoE blocks (#15720)
2e516f8 Daniel Han committed 24 days ago
AppImage E2E: count the app shell as rendered when the splash is already gone (#15764)
c61340c Daniel Han committed 24 days ago
CI: record the lazy CLI command import in the risky-loader baseline (#15762)
21127b2 Daniel Han committed 24 days ago
Studio: confirm Full Access directly from a chat tool approval (#15485)
3113197 Daniel Han committed 24 days ago
Bump the cargo-tauri group in /studio/src-tauri with 4 updates (#15360)
cef4c7f dependabot[bot] committed 24 days ago
Bump fast-uri from 3.1.8 to 3.1.9 in /docker/jupyter/unsloth_labext in the npm_and_yarn group across 1 directory (#15637)
1ad86b1 dependabot[bot] committed 24 days ago
Bump the pip group across 1 directory with 2 updates (#15518)
2630a52 dependabot[bot] committed 24 days ago
Studio frontend: bump zod to 4.6.7 (keep katex on 0.16) (#15746)
9869303 dependabot[bot] committed 24 days ago
Studio: train ChatML datasets whose roles are capitalized (#15599)
f787c47 Nilay committed 24 days ago
Studio login: say Unsloth and size the title with the UI font scale (#15759)
d4d0aef Daniel Han committed 24 days ago
Tests: the two gc-skip auto dist tests take their port from the shared helper (#15729)
2f2a431 Daniel Han committed 24 days ago
Studio llama-server args: a value-less ngram-mod flag keeps the next flag (#15728)
31377a6 Daniel Han committed 24 days ago
Studio fine-tuning queue: run the batch model probe behind the account's path guard (#15727)
70ff76c Daniel Han committed 24 days ago
Studio video: LTX-2.5 dense text-encoder fallback room check prices the load dtype (#14664)
17f6896 Daniel Han committed 24 days ago
Studio FLUX.3 upsampler: pin the sidecar to the native sd.cpp image card (#14696)
facc3a6 Daniel Han committed 24 days ago