Claude Sonnet 5.5 adds cyber safeguards and fallback to Sonnet 5

Anthropic released Claude Sonnet 5.5, the first Sonnet-tier model with cyber safeguards and model fallbacks like its flagship models. It scores 70.6% on Terminal-Bench 4.0 versus 66.4% for Opus 5.5 and achieved code execution in 178 of 410 ExploitBench runs. Blocked cyber requests fall back to Sonnet 5 in Anthropic's apps, but API developers must opt in.

• Sonnet 5.5 scores 70.6% on Terminal-Bench 4.0 vs 66.4% for Opus 5.5
• Achieved code execution in 178 of 410 ExploitBench runs
• CyScenarioBench: 46.1% vs 0.7% for Sonnet 5
• API fallback to Sonnet 5 must be enabled manually