Benchmark
SWE-bench Verified
Scaffold, tool permissions and time limit are part of the result; vendors run their own harnesses.
- Released
- 2024-08-13
- Categories
- coding_agent
- Openness
- unknown
- Source
- model_reports
- Reported scores
- 16
Reported scores
model_reports
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| Claude 3.7 Sonnet | Anthropic | 70.3 | 2025-02-24 | evidence |
| Claude 3.7 Sonnet | Anthropic | 63.7 | 2025-02-24 | evidence |
| Claude Opus 4 | Anthropic | 72.5 | 2025-05-22 | evidence |
| Claude Opus 5 | Anthropic | 96.0 | 2026-07-24 | evidence |
| Claude Sonnet 4 | Anthropic | 72.7 | 2025-05-22 | evidence |
| DeepSeek-R1 | DeepSeek | 49.2 | 2025-01-22 | evidence |
| DeepSeek-V3 | DeepSeek | 42.0 | 2024-12-27 | evidence |
| DeepSeek-V4-Flash | DeepSeek | 79.0 | 2026-06-25 | evidence |
| DeepSeek-V4-Pro | DeepSeek | 73.6 | 2026-04-22 | evidence |
| DeepSeek-V4-Pro | DeepSeek | 79.4 | 2026-04-22 | evidence |
| DeepSeek-V4-Pro | DeepSeek | 80.6 | 2026-04-22 | evidence |
| Gemini 2.5 Pro | 67.2 | 2025-06-17 | evidence | |
| Gemini 2.5 Pro | 59.6 | 2025-06-17 | evidence | |
| Gemini 3.1 Pro | 80.6 | 2026-02-19 | evidence | |
| GLM-5 | Z.ai | 77.8 | 2026-02-11 | evidence |
| Qwen3.5-397B-A17B | Qwen | 76.4 | 2026-02-16 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.