Benchmarks · coding
sudo L7
Surge AI's 60-task staff-level software engineering benchmark. Expert-authored rubrics assess correctness, architecture, production risk, verification and communication, using each model's published agent harness and reasoning setting.
15 models · 15 results · data as of 2026-10-29 · surgehq.ai ↗
Leaderboard
Opus 5.5 max
All results
15 rows| # | Model | Effort | sudo L7 | Harness |
|---|---|---|---|---|
| 1 | Opus 5.5 | max | 41.7% | Claude Code |
| 2 | Sonnet 5.5 | max | 41.7% | Claude Code |
| 3 | GPT-6 Astra | max | 33.9% | Codex |
| 4 | Fable 5.1 | max | 31.1% | Claude Code |
| 5 | GPT-6.1 Sol | max | 28.9% | Codex |
| 6 | Qwen 3.8 Max | xhigh | 20.6% | Qwen Code |
| 7 | Gemini 4 Argon | high | 18.9% | Antigravity |
| 8 | Muse Wonder (Max) | max | 15.6% | Muse Code |
| 9 | Grok 4.7 | xhigh | 15.6% | Grok Build |
| 10 | GLM 5.3 | max | 14.4% | OpenCode |
| 11 | Muse Spark 1.3 | max | 13.9% | Muse Code |
| 12 | Kimi K3 | max | 13.3% | OpenCode |
| 13 | Hy4 | high | 12.8% | OpenCode |
| 14 | Gemini 3.8 Flash | high | 11.7% | Antigravity |
| 15 | Mistral Large 4 | high | 11.7% | OpenCode |