EPIC-Bench - Ground Detection: leaderboard
Metric: Navigation score (0-100) on ground detection (localizing traversable ground regions), on EPIC-Bench (6,661 human-annotated image, text and mask tuples from 25 public datasets across 23 fine-grained embodied perception tasks; the model outputs bounding boxes, counts, path points or feasibility judgements that are scored against the masks), zero-shot, averaged over six runs for local models and two for API models; higher is better. Source: arxiv.org. Saturation forecast: Around September 2027. 89 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 397B A17B (Non-reasoning) | 51.94 |
| 2 | Qwen 3.5 122B A10B | 51.42 |
| 3 | Gemini 3 Flash (Preview) | 51.11 |
| 4 | Qwen 3.5 397B A17B | 51.1 |
| 5 | GPT-5.5 (Non-reasoning) | 50.97 |
| 6 | Qwen 3 VL 235B A22B Instruct | 50.96 |
| 7 | Qwen 3.5 Plus | 50.85 |
| 8 | Qwen 3.6 27B (Non-reasoning) | 49.99 |
| 9 | Qwen 3.5 27B (Non-reasoning) | 49.74 |
| 10 | Qwen 3.5 122B A10B (Non-reasoning) | 49.65 |
| 11 | Qwen 2.5 VL 72B Instruct | 49.45 |
| 12 | Qwen 3 VL 30B A3B Instruct | 49.37 |
| 13 | Gemini 3 Flash (Preview) (Non-reasoning) | 48.98 |
| 14 | Gemini 3 Pro | 48.82 |
| 15 | Qwen 3 VL 235B A22B (Thinking) | 48.18 |
Interactive version: theaggregate.ai/benchmark?slug=epic-bench-ground-detection · How It Works · Data refreshed daily, snapshot 2026-10-07.