Photo by Claude by Anthropic on November 24, 2025. May be a graphic of text that says '90 80 80.9% Software engineering SWE-bench Verified (n=500) 70 77.2% 60 74.5% 76.2% ၁ါ၀ 77.9% 50 76.3% 40 30 20 10 0 Opus4.5 4.5 Sonnet Sonnet4.5 4.5 Opus4.1 4.1 Gemini3Pro Pro GPT-5.1- GPT -5.1- Codex-Max GPT-5.1'.