EPIC-Bench - Spatial-Related Attributes: leaderboard

EPIC-Bench (2026) is a mask-grounded benchmark of embodied visual perception: 6,661 annotated image, text and mask samples across 23 tasks in target localization, navigation and manipulation, scored by localization, counting, path and feasibility metrics. This board scores grounding targets described by position, orientation and spatial relations, on a 0-100 scale; higher is better.

Metric: Target localization score (0-100) on the spatial-attribute tasks (targets described by position, orientation and spatial relations), 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 May 2028. 76 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3.1 Pro (Preview)56.65#102
2Qwen 3 VL 235B A22B (Thinking)55.67#462
3GPT-5.5Non-reasoning55.37#56
4Gemini 3 Pro54.5#146
5InternVL3.5-38B52.6#1026
6Qwen 3.5 Plus51.9#250
7Qwen 3.5 397B A17B51.82#270
8InternVL3.5-241B-A28B50.59#739
9Qwen 3.6 Plus50.17#245
10Qwen 3.5 27B50.03#380
11InternVL3.5-14B49.96#1519
12Doubao-Seed-1.6-vision49.92#366
13Qwen 2.5 VL 72B Instruct49.79#849
14Seed 1.849.68#258
15RoboBrain-2.0-32B49.67

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=epic-bench-spatial-related-attributes · How It Works · Data refreshed daily, snapshot 2026-11-22.