SWE-bench Verified (Bash Only): leaderboard
Bash-only variant of SWE-bench Verified for real-world GitHub issue resolution.
Metric: Score (%). Source: neosignal.io. Source: benchmarklist.com. Saturation forecast: Around April 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 | 64.8 | #83 |
| 2 | GPT-5.2 | 61.6 | #110 |
| 3 | Claude 3.7 Sonnet | 52.2 | #249 |
| 4 | DeepSeek V3 | 52.1 | #321 |
| 5 | O3 | 43.69 | #124 |
| 6 | GPT-4.1 | 41 | #248 |
| 7 | Grok 3 Mini | 38.6 | #281 |
| 8 | O4 Mini (Medium) | 34.6 | #180 (O4 Mini) |
| 9 | GPT-4.1 Mini | 32.8 | #355 |
| 10 | Qwen Plus | 28 | #385 |
| 11 | GPT-4o | 25.4 | #343 |
| 12 | Gemini 2.5 Pro (Preview 06-05) | 22 | #106 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=swe-bench-verified-bash-only · How It Works · Data refreshed daily, snapshot 2026-10-20.