PilotBench - Instruction Following: leaderboard
Metric: Instruction-following score (%): 0.5 x field completeness plus 0.3 x field validity plus 0.2 x format compliance of the predicted schema, 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 | 97.43 |
| 2 | Qwen 2.5 14B Instruct | 96.65 |
| 3 | Qwen 2.5 72B Instruct | 96.64 |
| 4 | glm-4-9B | 96.45 |
| 5 | DeepSeek V3 | 96 |
| 6 | GPT-4o Mini | 95.74 |
| 7 | Qwen 2.5 7B Instruct | 95.32 |
| 8 | Qwen 2.5 VL 72B Instruct | 94.89 |
| 9 | Qwen 2.5 32B Instruct | 94.66 |
| 10 | DeepSeek-R1-Distill-Qwen-7B | 94.42 |
| 11 | Qwen 3 14B | 94.09 |
| 12 | Doubao-1.5-Pro-32k | 93.5 |
| 13 | O3 Mini | 93.05 |
| 14 | DeepSeek R1 Distill Qwen 32B | 91.07 |
| 15 | QwQ-32B | 83.95 |
Interactive version: theaggregate.ai/benchmark?slug=pilotbench-instruction-following · How It Works · Data refreshed daily, snapshot 2026-10-07.