AgentDrive: 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 using Accuracy (%). Source: the original paper result table.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Around February 2027. 51 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | ChatGPT 4o | 82.5 | #296 |
| 2 | GPT-5 Chat | 81 | #250 |
| 3 | Qwen 3 235B A22B 2507 | 81 | #341 |
| 4 | Mistral Medium 3.1 | 80 | #418 |
| 5 | GPT-4.1 Mini | 77.5 | #363 |
| 6 | Qwen 3 Max | 77 | #213 |
| 7 | Gemini 2.5 Flash | 75.5 | #255 |
| 8 | GPT-4.1 | 75.5 | #258 |
| 9 | Phi-4-reasoning-plus | 75.5 | #1339 |
| 10 | GPT-4o (2024-11-20) | 75 | #392 |
| 11 | ERNIE 4.5 300B A47B | 75 | #478 |
| 12 | DeepSeek V3 (0324) | 73.5 | #349 |
| 13 | Hermes-3-Llama-3.1-405B | 71.5 | #531 |
| 14 | Llama 3.1 70B Instruct | 70.5 | #567 |
| 15 | Kimi K2 0905 | 69.5 | #292 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=agentdrive · How It Works · Data refreshed daily, snapshot 2026-11-05.