AgentDrive - Policy: leaderboard
Evaluates text reasoning about autonomous-driving scenarios on the 2,000-example AgentDrive-MCQ evaluation. Published model/version labels are preserved. This series reports Policy using Accuracy (%). Source: the original paper result table.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 51 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 Chat | 100 | #250 |
| 2 | ChatGPT 4o | 100 | #296 |
| 3 | GPT-4.1 Nano | 97.5 | #730 |
| 4 | Mistral Medium 3.1 | 97.5 | #418 |
| 5 | GPT-4.1 Mini | 95 | #363 |
| 6 | DeepSeek V3 (0324) | 95 | #349 |
| 7 | Qwen 3 Max | 95 | #213 |
| 8 | Claude Opus 4.1 | 95 | #146 |
| 9 | ERNIE 4.5 300B A47B | 95 | #478 |
| 10 | Gemini 2.5 Flash | 92.5 | #255 |
| 11 | Phi-4-reasoning-plus | 92.5 | #1339 |
| 12 | GPT-4.1 | 90 | #258 |
| 13 | Kimi K2 | 90 | #252 |
| 14 | Nova Micro (v1) | 90 | #840 |
| 15 | Grok 3 Mini | 87.5 | #289 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=agentdrive-policy · How It Works · Data refreshed daily, snapshot 2026-11-05.