Coding
SWE-Bench Verified
Real coding tasks pulled from open-source projects — the AI has to find and fix actual bugs. A human-checked version of the original SWE-Bench. Higher is better.
Rankings
Higher is better195.5%
Claude Fable 5
Anthropic · Jun 9 2026
287.6%
Claude Opus 4.7
Anthropic · Apr 16 2026
380.9%
Claude Opus 4.5
Anthropic · Nov 24 2025
480.8%
Claude Opus 4.6
Anthropic · Feb 5 2026
580.6%
Gemini 3.1 Pro
Google · Feb 19 2026
680.2%
Kimi K2.6
Moonshot AI · Apr 21 2026
780%
GPT-5.2
OpenAI · Dec 11 2025
879.6%
Claude Sonnet 4.6
Anthropic · Feb 17 2026
978%
Gemini 3.0 Flash
Google · Dec 17 2025
1077.8%
GLM-5
Z.ai · Feb 12 2026
1177.6%
Mistral Medium 3.5
Mistral · Apr 29 2026
1277.4%
Muse Spark
Meta · Apr 8 2026
1377.2%
Claude Sonnet 4.5
Anthropic · Sep 29 2025
1476.8%
Kimi K2.5
Moonshot AI · Jan 27 2026
1576.3%
GPT-5.1
OpenAI · Nov 12 2025
1676.2%
Gemini 3.0 Pro
Google · Nov 18 2025
1774.5%
Claude Opus 4.1
Anthropic · Aug 5 2025
1873.8%
GLM-4.7
Z.ai · Dec 22 2025
1973.3%
Claude Haiku 4.5
Anthropic · Oct 15 2025
2072.7%
Claude Sonnet 4
Anthropic · May 22 2025
2172.5%
Claude Opus 4
Anthropic · May 22 2025
2271.3%
Kimi K2 Thinking
Moonshot AI · Nov 6 2025
2368%
GLM-4.6
Z.ai · Sep 30 2025
2465.8%
Kimi K2
Moonshot AI · Jul 11 2025
2465.8%
Kimi K2 (0905)
Moonshot AI · Sep 5 2025
2664.2%
GLM-4.5
Z.ai · Jul 28 2025
2762.3%
Claude 3.7 Sonnet
Anthropic · Feb 24 2025
2859.8%
GLM-4.5-Air
Z.ai · Jul 28 2025
2959.6%
Gemini 2.5 Pro
Google · Mar 25 2025
3049%
Claude 3.5 Sonnet (upgraded)
Anthropic · Oct 22 2024
3140.6%
Claude 3.5 Haiku
Anthropic · Oct 22 2024
3233.4%
Claude 3.5 Sonnet
Anthropic · Jun 20 2024
3333%
Claude 3 Opus
Anthropic · Mar 4 2024