Last released Apr 27, 2026
Chain-of-thought monitorability and faithfulness evaluation for reasoning-model agents.
Supported by