SWE-bench Verified

Human-verified subset of SWE-bench. Tests whether a model can resolve real GitHub issues by editing code.

Category: coding · Metric: pass@1 · Source: www.swebench.com ↗

Leaderboard

RankModelProviderScoreMeasuredSource
1 Claude Opus 4.7 Anthropic 77.2 2026-04-22
2 GPT-5 OpenAI 74.9 2025-08-07
3 Claude Sonnet 4.6 Anthropic 72.5 2026-03-12
4 o3 OpenAI 71.7 2025-04-16
5 Qwen3-Coder-480B Alibaba 69.6 2025-07-22
6 Gemini 2.5 Pro Google 63.8 2025-03-25

What this benchmark measures, in detail

Human-verified subset of SWE-bench. Tests whether a model can resolve real GitHub issues by editing code.

Different benchmarks measure different things. A model that excels on SWE-bench Verified may underperform on real-world workloads if the benchmark's distribution doesn't match your data. Use benchmark scores as a triage signal — narrow to a shortlist — then evaluate on your actual workload before committing.

Methodology notes

Scores in the leaderboard are taken from the model's release announcement or model card, cited via the "Source" link. Where two sources disagree (which happens often for SWE-bench and IFEval), the linked primary source wins. Reproducibility for some benchmarks (notably anything graded by an LLM) varies by run — treat the score as ±2-3 points unless the source is a peer-reviewed result.

Related benchmarks