GitTaskBench (SWE-Agent) - Execution Completion: leaderboard
GitTaskBench execution completion rate (Ni et al., 2025): the percentage of the 54 repository-driven tasks for which the agent runs the repository and produces a non-empty output in the expected format, whether or not it passes, under SWE-agent. Averaged over two runs; higher is better.
Metric: Execution completion rate (%). Source: gittaskbench.github.io. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3.7 Sonnet (20250219) | 64.81 | #399 |
| 2 | Claude 3.5 Sonnet (20241022) | 41.67 | #635 |
| 3 | GPT-4.1 | 38.89 | #492 |
| 4 | O3 Mini (Medium) | 25.93 | #536 (O3 Mini) |
| 5 | Llama 3.3 70B Instruct | 25.83 | #1353 |
| 6 | DeepSeek V3 (0324) | 18.52 | #704 |
| 7 | GPT-4o (2024-08-06) | 17.58 | #731 |
| 8 | Qwen 3 32B (Thinking) | 16.67 | #960 (Qwen 3 32B) |
| 9 | Qwen 3 32B (Non-reasoning) | 7.41 | #960 (Qwen 3 32B) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=gittaskbench-swe-agent-execution-completion · How It Works · Data refreshed daily, snapshot 2026-11-12.