www.anthropic.com
网页
综合
Claude SWE-Bench performance \ Anthropic
Our latest model, the upgraded Claude 3.5 Sonnet, achieved 49% on SWE-bench Verified, a software engineering evaluation, beating the previous state-of-the-art model's 45%. This post explains…
2025-01-06
自动蜘蛛
相关度 6.8