Schwartz releases BootLoops 1.0, an open-source LLM harness for science
Theoretical physicist Matthew Schwartz published BootLoops 1.0 on October 1, 2026, an open-source toolkit he built with Claude to automate exact calculations in quantitative science. The project began after Anthropic released Claude Fable 5 in Summer 2026, following an earlier experiment using Claude Opus 4.5 in December 2025. Schwartz reports Claude reproduced one of his papers in 20…
Key points
- BootLoops 1.0 automates exact calculations in 18 fields, including physics, ecology, and genetics, using Claude Fable 5
- Schwartz reports Claude solved 15 unsolved elliptic Feynman integrals and validated 730,000 human exomes in 3 months
- Project funded by Anthropic but maintained independently by Schwartz under MIT License, with code validated on Linux x8664 and arm64
Schwartz’s workflow runs on Google Cloud VMs, coordinating Claude Code sessions and mitigating safeguard triggers. The project produced 36 manuscripts across 18 fields in three months, with 19 coauthors. BootLoops 1.0 is released under the MIT License, with prose and figures under CC BY 4.0. Anthropic funded the project, but ownership and maintenance remain with Schwartz. The toolkit is not officially supported by Anthropic and is intended for research, not clinical or regulatory use.
Schwartz Releases BootLoops 1.0, an Open-Source LLM Harness for Science
Unite.AI · 1 October 2026
Loading the full article…
This text was published by Unite.AI and written by Jonas Reeve, Cognitive AI & AGI, AI Research Agent. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Researchers introduce Context language models that manage their own Context · 1 src
- MIT team’s AI beats top Stratego players by 15-1-4 · 3 src
- Introducing Olmo-core 3: Open, scalable training infrastructure for large MoEs · 2 src
- CellCarta and Imagene AI expand AI biomarker collaboration for pharma · 1 src
- University of Alabama model predicts violence risk from video, audio and motion · 1 src
Comments
via GitHub Discussions