Back to Benchmarks
💻
SWE-bench V
SWE-bench Verified
500 real GitHub issue fixes; gold standard for production coding
💻 Coding
Best: 88.6%
18 Models Evaluated
Champion
Claude Opus 4.8
Anthropic
88.6%
Best Score
Model Rankings
All models ranked by score on SWE-bench V. Scores are relative to the best score of 88.6%.
| # | Model | Provider | Score | vs Best |
|---|---|---|---|---|
| 1 | Claude Opus 4.8 | Anthropic | 88.6% | Best |
| 2 | Claude Opus 4.7 Elo 1500 | Anthropic | 87.6% | -1.0% |
| 3 | Claude Opus 4.6 Elo 1503 | Anthropic | 80.8% | -7.8% |
| 4 | DeepSeek V4 Pro (Max) | DeepSeek | 80.6% | -8.0% |
| 5 | MiniMax M2.5 Elo 1404 | MiniMax | 80.2% | -8.4% |
| 6 | Claude Sonnet 4.6 Elo 1460 | Anthropic | 79.6% | -9.0% |
| 7 | Gemini 3.1 Pro Elo 1492 | 78% | -10.6% | |
| 8 | GLM-5 Elo 1454 | Zhipu AI | 77.8% | -10.8% |
| 9 | Kimi K2.5 Elo 1438 | Moonshot | 76.8% | -11.8% |
| 10 | Qwen 3.5 Elo 1450 | Alibaba | 76.4% | -12.2% |
| 11 | Gemini 3 Pro Elo 1486 | 75% | -13.6% | |
| 12 | Step-3.5-Flash Elo 1389 | Stepfun | 74.4% | -14.2% |
| 13 | MiMo-V2-Flash Elo 1393 | Xiaomi | 73.4% | -15.2% |
| 14 | DeepSeek V3.2 Elo 1423 | DeepSeek | 67.8% | -20.8% |
| 15 | GPT-oss 120B Elo 1355 | OpenAI | 62.4% | -26.2% |
| 16 | GPT-5.5 Elo 1478 | OpenAI | 58.6% | -30.0% |
| 17 | DeepSeek R1 Elo 1398 | DeepSeek | 49.2% | -39.4% |
| 18 | Grok 3 Elo 1412 | xAI | 49% | -39.6% |
Score Distribution
Claude Opus 4.8
88.6%
Claude Opus 4.7
87.6%
Claude Opus 4.6
80.8%
DeepSeek V4 Pro (Max)
80.6%
MiniMax M2.5
80.2%
Claude Sonnet 4.6
79.6%
Gemini 3.1 Pro
78%
GLM-5
77.8%
Kimi K2.5
76.8%
Qwen 3.5
76.4%
Run This Benchmark
Learn how to evaluate your own models on SWE-bench V.
SWE-bench V is a standardized benchmark for evaluating AI model capabilities.
To run this benchmark on your own models, visit the official benchmark page or leaderboard for detailed instructions, code, and evaluation scripts.
Data compiled from official model cards, public leaderboards, and independent evaluations.