RideWay: leaderboard
Metric: Efficiency Utility (success-gated, held-out custom14 mean over 14 tasks x 3 attempts; a successful trajectory is discounted for excess tool calls and user-facing turns beyond human reference floors, penalties fit from human pairwise preferences; failures score 0). Source: arxiv.org. Saturation forecast: Around September 2027. 24 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 (Non-reasoning) | 70.5 |
| 2 | Qwen 3.7 Max (Non-reasoning) | 66.1 |
| 3 | Qwen 3.6 27B (Non-reasoning) | 63.8 |
| 4 | Claude Sonnet 5 | 63.7 |
| 5 | Gemma 4 31B | 62.8 |
| 6 | GLM-5.2 (Non-reasoning) | 62 |
| 7 | Qwen 3.6 Plus (Non-reasoning) | 61.2 |
| 8 | Claude Opus 4.6 | 60.6 |
| 9 | MiMo-V2.5-Pro (Non-reasoning) | 59 |
| 10 | Kimi K2.6 (Non-reasoning) | 56.3 |
| 11 | DeepSeek V4 Pro (Non-reasoning) | 54.3 |
| 12 | Gemma 4 26B | 49.3 |
| 13 | Gemini 3.5 Flash (Minimal) | 44.3 |
| 14 | Qwen 3.5 9B (Non-reasoning) | 43.1 |
| 15 | Qwen 3.6 35B A3B (Non-reasoning) | 32.3 |
Interactive version: theaggregate.ai/benchmark?slug=rideway · How It Works · Data refreshed daily, snapshot 2026-09-26.