Safely scaling multi-agent teams with Claude Fable 5.2
In a toy experiment, a team of 1,000 agents running an early version of Claude Fable 5.2 successfully reproduced a formal proof of Fermat's Last Theorem using a shared SQL database, Git, general-purpose tools, and layered safety controls without human steering. The experiment demonstrated that large multi-agent teams can tackle ambitious, long-horizon tasks when provided with carefully designed safety boundaries and robust coordination primitives. Open article →
- Anthropic's Fable 5.2 formalized Fermat's Last Theorem in just 30 hours - A team of approximatel...
- 1,000 Claude Fable 5.2 agents formalized Fermat's Last Theorem in 30 hours Anthropic's first for...
- hold on. anthropic says 1,000 fable 5.2 agents formalized the existing proof of fermat’s last the...
- Holy Shit this is AGI A prototype operating system, built from scratch as an experiment Workin...
- This is INSANE. Anthropic says one of its engineers asked Claude Fable 5.2 to build an operating ...
- The model is the most important thing. Opus 5.5 with 1000 agents completed only 26% of the Fermat...
- this is actually insane Fable 5.2 coordinated over 1,000 AI agents for 8 days to build an operat...
- "A team of 1,000 Fable 5.2 agents fully formalized Fermat’s Last Theorem in 30 hours" Yeah... Th...
- i'm happy to share some recent work on safely scaling a team of 1,000 Fable 5.2 agent to tackle a...
- Imagine every scientist on Earth having access to not just 1000s, but millions of AI scientists a...
- Claude Fable 5.2 just built a fully bootable operating system from an empty folder. coordinated ...
- > Anthropic engineer gives Claude Fable 5.2 an empty folder and asks for an operating system > it...