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.
17 models · 17 results · data as of 2026-11-06 · surgehq.ai ↗
Leaderboard
Sonnet 5.5 max
All results
17 rows| # | Model | Effort | sudo L7 |
|---|---|---|---|
| 1 | Sonnet 5.5 | max | 50.6% |
| 2 | Opus 5.5 | max | 45.6% |
| 3 | GPT-6.1 Sol | max | 33.9% |
| 4 | Fable 5.1 | max | 32.2% |
| 5 | GPT-6 Astra | max | 31.1% |
| 6 | Kimi K3.1 | max | 16.7% |
| 7 | Qwen 3.8 Max | xhigh | 16.7% |
| 8 | Muse Wonder (Max) | max | 16.1% |
| 9 | Gemini 4 Argon | high | 16.1% |
| 10 | Hy4 preview | high | 15.6% |
| 11 | Inkling 2 Small | xhigh | 14.1% |
| 12 | Mistral Large 4 | high | 13.9% |
| 13 | Muse Spark 1.3 | max | 12.8% |
| 14 | Grok 4.7 | xhigh | 12.2% |
| 15 | Kimi K3 | max | 12.2% |
| 16 | Gemini 3.8 Flash | high | 10.6% |
| 17 | GLM 5.3 | max | 9.4% |