Benchmarks · Professional work · Agents & computer use · Science & research
RWE-Bench
Real engineering design work: 20 tasks in electrical, aerospace and mechanical engineering, each run up to 5 times for up to 24 hours. A run passes when its design passes every check; the score is the average pass rate. From Xplore.
15 models · 15 results · updated 2026-11-12 · xplore.ai ↗
Leaderboard
Fable 5.2 max
Heavy line: best score at each price. Thin lines join one model's effort levels. Cost uses each model's median subscription plan, or API price with no plan data. 5 more models under All models in settings.
Pass at k
As published by xplore.ai, one line per run · follows the leaderboard’s model and effort settings
Top 10 of 15 runs by score ·
More from this benchmark
2 published measuresTime per task15 results · lower is better
5 more models under All models in settings.
Score vs any published measure1 measures: sub-scores, splits, counts
All results
15 rows| # | Model | Effort | RWE-Bench | Subscription $ / task | API $ / task | Time | Harness |
|---|---|---|---|---|---|---|---|
| 1 | Fable 5.2 | max | 62.0% ±6.5 | $4.22 | $115 | 5.5 h | Claude Code |
| 2 | GPT-6 Astra | max | 37.0% ±6.5 | $2.52 | $59 | 2.8 h | Codex |
| 3 | Opus 5.5 | max | 29.0% ±7.9 | $0.87 | $58 | 4.2 h | Claude Code |
| 4 | GPT-6.1 Sol | max | 25.0% ±6.5 | $0.34 | $10 | 3.2 h | Codex |
| 5 | Gemini 4 Carbon | high | 20.0% ±6.4 | $22 | $31 | 7.1 h | Antigravity |
| 6 | GPT-6.1 Terra | max | 19.0% ±5.8 | $0.44 | $11 | 2.7 h | Codex |
| 7 | Gemini 4 Argon | high | 12.0% ±5.5 | $17 | $24 | 2.5 h | Gemini CLI |
| 8 | Fable 5.1 | max | 7.0% ±4.3 | $15 | $98 | 3.7 h | Claude Code |
| 9 | Kimi K3.1 | max | 1.0% ±2.3 | $71 | $71 | 4.1 h | Kimi Code |
| 10 | Muse Wonder | max | 0.0% ±1.1 | $18 | $28 | 2.7 h | Muse Code |
| 11 | Nemotron 3.5 Super VL | max | 0.0% ±1.1 | free | $2.93 | 7.6 h | OpenHands |
| 12 | DeepSeek V4.1 Flash | max | 0.0% ±1.1 | $0.074 | $1.03 | 2.0 h | DSH |
| 13 | GLM 5.3 | max | 0.0% ±1.1 | $2.35 | $23 | 4.6 h | ZCode |
| 14 | Grok 4.7 | xhigh | 0.0% ±1.1 | $6.15 | $66 | 3.7 h | Grok Build |
| 15 | Qwen 3.8 Max | xhigh | 0.0% ±1.1 | $2.06 | $17 | 4.4 h | Qwen Coder |