DriveBench: leaderboard
Autonomous-driving VLM reliability benchmark across clean, corrupted, and text-only inputs for perception, prediction, planning, and behavior tasks.
Metric: Average Accuracy (%). Source: github.com. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 51.79 | #343 |
| 2 | Qwen 2 VL 7B | 42.66 | #750 |
| 3 | Qwen 2 VL 72B | 40.95 | #460 |
| 4 | InternVL2-8B | 40.8 | #817 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=drivebench · How It Works · Data refreshed daily, snapshot 2026-10-20.