WildRoadBench — leaderboard
Metric: AP50 (self-reported). Source: benchmarklist.com. 25 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.6 Plus | 36.8 |
| 2 | Gemini 3.1 Flash Lite | 25.8 |
| 3 | Claude Sonnet 4.6 | 25 |
| 4 | Qwen 2.5 VL 7B Instruct | 23.6 |
| 5 | Qwen 3 VL 8B Instruct | 21.3 |
| 6 | GPT-5.4 | 20.9 |
| 7 | Gemini 2.5 Flash | 16.3 |
| 8 | Qwen 3 VL 8B (Thinking) | 14.8 |
| 9 | Claude Opus 4.6 | 12.7 |
| 10 | InternVL3-8B | 10.9 |
| 11 | GPT-4o | 9.5 |
| 12 | InternVL3-14B | 7.9 |
| 13 | InternVL2-8B | 2.2 |
Interactive version: theaggregate.ai/benchmark?slug=wildroadbench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.