AgentDrive - Comparative: 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 Comparative using Accuracy (%). Source: the original paper result table.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Not forecast. 51 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o (2024-11-20) | 95 | #392 |
| 2 | Qwen 3 Max | 95 | #213 |
| 3 | Mistral Medium 3.1 | 95 | #418 |
| 4 | Phi-4-reasoning-plus | 95 | #1339 |
| 5 | Gemini 2.5 Flash | 92.5 | #255 |
| 6 | Gemma 3 27B (IT) | 92.5 | #528 |
| 7 | Llama 3.1 70B Instruct | 92.5 | #567 |
| 8 | Claude Opus 4.1 | 92.5 | #146 |
| 9 | Claude 3.7 Sonnet (Thinking) | 92.5 | #259 (Claude 3.7 Sonnet) |
| 10 | GPT-5 Chat | 92.5 | #250 |
| 11 | Qwen 3 235B A22B 2507 | 92.5 | #341 |
| 12 | Llama 3.3 70B Instruct | 90 | #541 |
| 13 | GPT-4.1 Mini | 90 | #363 |
| 14 | GPT-4.1 | 90 | #258 |
| 15 | ChatGPT 4o | 90 | #296 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=agentdrive-comparative · How It Works · Data refreshed daily, snapshot 2026-11-05.