Evaluation

Frontier risk evaluations (CBRN, cyber, autonomy) for upgraded Claude 3.5 Sonnet / 3.5 Haiku

automated CBRN knowledge tests and non-expert uplift; CTF challenges (pwn, reverse engineering, cryptography, web, network); software-engineering tasks (PR satisfying test requirements)

Release & deployment checks 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 3.5 Haiku, Claude 3.5 Sonnet (New) Oct 2024

Anthropic

rich detail 22 method fields published

See how this test was run