EPIC-Bench - Feasible Path: leaderboard
Metric: Navigation path score (0-100) on feasible path planning (a point path from the camera or a marked region to the target that stays on traversable ground), 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 January 2027. 89 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro | 70.82 |
| 2 | Gemini 3.1 Pro (Preview) | 70.32 |
| 3 | Claude Sonnet 4.6 (Thinking) | 68.46 |
| 4 | Qwen 3.5 122B A10B | 63.47 |
| 5 | Qwen 3.6 Plus | 62.12 |
| 6 | Qwen 3.5 35B A3B | 61.17 |
| 7 | Qwen 3.5 397B A17B | 60.11 |
| 8 | Qwen 3.6 27B (Non-reasoning) | 59.15 |
| 9 | Qwen 3.5 Plus | 57.89 |
| 10 | Qwen 3.5 27B (Non-reasoning) | 53.5 |
| 11 | Seed 1.8 | 53.43 |
| 12 | Gemini 3 Flash (Preview) | 52.63 |
| 13 | Gemini 2.5 Flash (Non-reasoning) | 51.98 |
| 14 | Qwen 3 VL 235B A22B (Thinking) | 51.19 |
| 15 | GLM-4.6V | 51.11 |
Interactive version: theaggregate.ai/benchmark?slug=epic-bench-feasible-path · How It Works · Data refreshed daily, snapshot 2026-10-07.