EPIC-Bench - Visual Matching: 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 finding an object marked in one view in another view of the same scene, on a 0-100 scale; higher is better.

Metric: Navigation score (0-100) on visual matching (localizing an object marked in one view in another view of the same scene), 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 June 2027. 76 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3 Pro60.08#145
2Gemini 3.1 Pro (Preview)59.06#99
3GPT-5.5Non-reasoning51.65#54
4Gemini 3 Flash (Preview)49.92#152
5Qwen 3 VL 235B A22B (Thinking)49.01#459
6Qwen 3.5 27B46.77#375
7GPT-5 Mini45.97#364
8Seed 1.845.9#252
9O4 Mini45.63#368
10Qwen 3.5 397B A17B45.35#264
11Qwen 3.5 Plus45.21#242
12Qwen 3.5 122B A10B45.19#346
13GPT-5.1Non-reasoning44.69#240
14GPT-5.4Non-reasoning44.35#135
15GPT-5.2Non-reasoning44.1#185

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=epic-bench-visual-matching · How It Works · Data refreshed daily, snapshot 2026-11-21.