ENC-Bench: leaderboard
Metric: Mean accuracy (%) over ENC-Bench's seven multiple-choice perception and maritime decision-making tasks (symbol recognition; point, linestring and polygon features; track direction, safety passage and anchorage selection) 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; the spatial reasoning tasks are not included; higher is better. Source: arxiv.org. Saturation forecast: Around March 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 47.88 | #145 |
| 2 | Qwen 3 VL 235B A22B Instruct | 46.91 | #264 |
| 3 | Qwen 3 VL 235B A22B (Thinking) | 46.68 | #228 (Qwen 3 VL 235B A22B) |
| 4 | Gemini 2.5 Flash | 44.85 | #237 |
| 5 | Qwen 3 VL 32B Instruct | 42.82 | #276 |
| 6 | Qwen 3 VL 32B (Thinking) | 40.59 | #287 (Qwen 3 VL 32B) |
| 7 | GLM-4.5V | 38.55 | #339 |
| 8 | InternVL3-38B | 37.04 | #395 |
| 9 | Llama 4 Maverick Instruct | 36.44 | #439 |
| 10 | GPT-4o | 34.87 | #333 |
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 · How It Works · Data refreshed daily, snapshot 2026-10-11.