Benchmarks · Agents & computer use · Science & research · Coding
OpenWorldEngineer
Uses Mercor's overall mean: average attempts within each task, then weight each scored task equally. Score: mean normalized score across scored tasks. Collected by Jake Boggs' benchmark dashboard.
14 models · 14 results · data as of 2026-11-15 · openworldengineer.mercor.com ↗
Leaderboard
Fable 5.2 max
All results
14 rows| # | Model | Effort | OpenWorldEngineer |
|---|---|---|---|
| 1 | Fable 5.2 | max | 54.6% |
| 2 | GPT-6 Astra | max | 44.2% |
| 3 | Opus 5.5 | max | 41.6% |
| 4 | GPT-6.1 Sol | max | 41.2% |
| 5 | Sonnet 5.5 | max | 39.0% |
| 6 | Gemini 4 Carbon | high | 32.8% |
| 7 | Fable 5.1 | max | 29.1% |
| 8 | Gemini 4 Argon | high | 24.1% |
| 9 | Kimi K3.1 | max | 13.6% |
| 10 | Muse Wonder | max | 12.7% |
| 11 | DeepSeek V4.1 Flash | max | 12.4% |
| 12 | GLM 5.3 | max | 11.8% |
| 13 | Grok 4.7 | xhigh | 7.0% |
| 14 | Inkling | max | 1.1% |