ENC-Bench - Bearing Calculation: leaderboard
Metric: Acc@20 degrees accuracy (%): share of ENC-Bench spatial reasoning answers with a computed true bearing between two marked points within 20 degrees of the truth (1,161 questions), numeric answers on ENC-Bench's 840 NOAA S-57 electronic navigational charts rendered with OpenCPN in day, dusk and night palettes at six scales, zero-shot; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 235B A22B (Thinking) | 55.64 | #228 (Qwen 3 VL 235B A22B) |
| 2 | Gemini 2.5 Pro | 46.86 | #145 |
| 3 | Qwen 3 VL 32B (Thinking) | 44.1 | #287 (Qwen 3 VL 32B) |
| 4 | InternVL3-38B | 41.09 | #395 |
| 5 | Qwen 3 VL 235B A22B Instruct | 36.95 | #264 |
| 6 | Gemini 2.5 Flash | 28.77 | #237 |
| 7 | GLM-4.5V | 28.6 | #339 |
| 8 | Qwen 3 VL 32B Instruct | 26.44 | #276 |
| 9 | GPT-4o | 25.15 | #333 |
| 10 | Llama 4 Maverick Instruct | 14.9 | #439 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=enc-bench-bearing-calculation · How It Works · Data refreshed daily, snapshot 2026-10-11.