PilotBench: leaderboard
Metric: Pilot-Score (5-100): 0.6 x regression score plus 0.4 x instruction-following score, on PilotBench (one-step trajectory and attitude prediction from 34-channel telemetry of 708 general-aviation trajectories in nine flight phases, fixed output schema, temperature 0); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 19 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 32B | 90.23 |
| 2 | Qwen 2.5 72B Instruct | 88.58 |
| 3 | DeepSeek V3 | 88.31 |
| 4 | GPT-4o Mini | 88.09 |
| 5 | O3 Mini | 87.76 |
| 6 | Doubao-1.5-Pro-32k | 87.52 |
| 7 | Qwen 2.5 32B Instruct | 85.27 |
| 8 | QwQ-32B | 84.9 |
| 9 | Qwen 2.5 14B Instruct | 84.44 |
| 10 | Qwen 2.5 VL 72B Instruct | 77.35 |
| 11 | Qwen 2.5 7B Instruct | 71.11 |
| 12 | glm-4-9B | 69.03 |
| 13 | DeepSeek R1 Distill Llama 70B | 65 |
| 14 | DeepSeek R1 Distill Qwen 32B | 59.37 |
| 15 | Qwen 3 14B | 49.54 |
Interactive version: theaggregate.ai/benchmark?slug=pilotbench · How It Works · Data refreshed daily, snapshot 2026-10-07.