Claude 3.7 Sonnet tops Aider Polyglot leaderboard with 32k thinking tokens

Claude 3.7 Sonnet has taken the top place on Paul Gauthier's Aider Polyglot benchmark, an independent evaluation focused on code capabilities.

• The model achieved first place when run with an increased 32,000 thinking token limit.
• It spent $36.83 running the entire benchmark, which is significantly more than the previous leader but less than third-place o1-high.
• No results are yet available for Claude 3.7 Sonnet on the LM Arena leaderboard.