We did many new improvements and fixes to Studio!
- Tool calls for all models are now +30% to +80% more accurate.
- Web search now actually gets web content and not just summaries
- Number of tool calls allowed are increased to 25 from 10
- Tool calls now terminate much better, so looping / repetitions will be reduced
- More tool call healing and de-duplication logic to stop tool callings from leaking XML as well
- Tested with
unsloth/Qwen3.5-4B-GGUF(UD-Q4_K_XL), web search + code execution + thinking enabled.
| Metric | Before | After |
|---|---|---|
| XML leaks in response | 10/10 | 0/10 |
| URL fetches used | 0 | 4/10 runs |
| Runs with correct song names | 0/10 | 2/10 |
| Avg tool calls | 5.5 | 3.8 |
| Avg response time | 12.3s | 9.8s |
New features
- Update button now visible
- Install script styling all updated!
- Added custom folders so you can use any GGUFs in any folder - for now access in Advanced Settings in Chat and Custom Folders
- Preliminary Automatic Multi GPU support for inference and training - useful for large models that don't fit on 1 GPU - Studio auto will allocate GPU resources
- Intel Macs should work out of the box
Much smoother and faster Studio
- Fixed timeouts of downloads of large models - no more timeouts seen.
- Fixed Hugging Face rate limiting - HF API calls reduced by 90%
- Fixed bun on Windows and faster installs
To update Studio:
- For Linux, WSL, Mac, do:
unsloth studio update - For Windows native, do:
irm https://unsloth.ai/install.ps1 | iex - For Linux, WSL, Mac reinstalls, do:
curl -fsSL https://unsloth.ai/install.sh | sh
What's Changed
- Fix LM Studio GGUF loading on native Windows (no GPU) by @danielhanchen in #4665
- studio: add HF/local model selection UI for GGUF export by @Imagineer99 in #4365
- Fix blank page on Windows due to broken .js MIME type by @rolandtannous in #4674
- fix: [Studio] setup.ps1 update-flow for windows by @rolandtannous in #4667
- studio: unify Windows installer/setup logging style, verbosity controls, and startup messaging by @Imagineer99 in #4651
- studio: preserve GGUF context max after apply and refresh by @Imagineer99 in #4691
- [Studio] multi gpu finetuning/inference via "balanced_low0/sequential" device_map by @Datta0 in #4602
- Fix editable install scanning 6,500+ node_modules dirs by @danielhanchen in #4697
- fix(studio): avoid UnicodeEncodeError on Windows cp1252 consoles by @danielhanchen in #4699
- Fix/bun windows bin detection by @Etherll in #4703
- fix: skip download progress polling for exported GGUF models by @rolandtannous in #4709
- [Studio] Fix: replace hard timeout with inactivity timeout for model loading by @rolandtannous in #4707
- [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci[bot] in #4705
- studio: prevent false multimodal warning during model loading by @Imagineer99 in #4704
- fix(studio): open tour ReadMore links in new tab by @danielhanchen in #4694
- [studio] multi gpu: revert to balanced for inference. by @Datta0 in #4698
- fix: throttle and cache HuggingFace modelInfo API calls by @Shine1i in #4696
- fix(studio): correct default weight_decay and learning rate by @danielhanchen in #4695
- fix: auto-retry stalled HF downloads with HF_HUB_DISABLE_XET=1 by @rolandtannous in #4712
- studio: add update button to navbar with guided commands and cross-platform support by @Imagineer99 in #4721
- studio: improve GGUF tool calling accuracy and reliability by @danielhanchen in #4700
- studio: fix export HF model dropdown clearing on enter/click-away by @Imagineer99 in #4726
- Studio: simplify tool-call dedup and replace html2text with builtin converter by @danielhanchen in #4722
- feat: custom scan folders for GGUF model discovery by @Shine1i in #4723
- Bump installer minimum version pin to 2026.3.18 by @danielhanchen in #4729
Full Changelog: v0.1.25-beta...v0.1.3-beta