Skip to content
Inside the Frontier

Evaluation

Text-based reinforcement learning task

ability to develop scaffolding (e.g. ReACT, Tree of Thought) to significantly enhance a weaker model's performance on a text-based RL task (proxy for self-optimization)

Autonomy & self-improvement Anthropic
Ask about this evaluation

How it was run

Open a release to see the setup, scoring, and other details that the lab published. Only fields the lab actually disclosed are shown.

Claude Opus 4, Claude Sonnet 4
Anthropic

rich detail 23 method fields published

See how this test was run