Coding Agent Leaderboard - SWE-Bench Pro -- Ansible: leaderboard
Metric: Score (%). Source: huggingface.co. 24 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Opus 4.8 + OpenCode | 78.1 |
| 2 | Opus 4.8 + Claude Code | 69.8 |
| 3 | GPT 5.5 - high + Codex | 60.4 |
| 4 | Qwen3.6-27B-FP8 + OpenCode | 57.3 |
| 5 | Qwen3.6-27B-FP8 + Claude Code | 52.1 |
| 6 | Sonnet 4.6 + Claude Code | 50 |
| 7 | Qwen3.6-27B-FP8 + Pi | 49 |
| 8 | Qwen3.6-35B-A3B-NVFP4 + Pi | 47.9 |
| 9 | gemma-4-31B-it-FP8-block + Pi | 46.9 |
| 10 | Qwen3.6-35B-A3B-NVFP4 + Claude Code | 45.8 |
Interactive version: theaggregate.ai/benchmark?slug=coding-agent-leaderboard-swe-bench-pro-ansible · How It Works · Data refreshed daily, snapshot 2026-09-05.