AeroCopilotBench - Tier-1 Aviation Knowledge: leaderboard
AeroCopilotBench (arXiv 2608.16349v1, Table 4) Tier-1: 1,200 aviation knowledge multiple-choice questions from authoritative sources, graded by exact answer letter, for the six models evaluated on both tiers, thinking enabled at high effort; accuracy percent, higher is better.
Metric: Accuracy (%; 1,200 multiple-choice questions). Source: arxiv.org. Saturation forecast: Estimated already saturated. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.7 Max (Thinking) | 89.17 | #85 (Qwen 3.7 Max) |
| 2 | Qwen 3.5 397B A17B (Thinking) | 86.33 | #154 (Qwen 3.5 397B A17B) |
| 3 | DeepSeek V4 Pro (High) | 85.92 | #112 (DeepSeek V4 Pro) |
| 4 | DeepSeek V4 Flash (High) | 83 | #155 (DeepSeek V4 Flash) |
| 5 | Qwen 3.5 122B A10B (Thinking) | 82.5 | #184 (Qwen 3.5 122B A10B) |
| 6 | MiniMax-M2.5 | 74.42 | #316 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=aerocopilotbench-tier-1-aviation-knowledge · How It Works · Data refreshed daily, snapshot 2026-11-05.