EPIC-Bench - Placement Region: 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 where an object can be placed, with a physical feasibility judgement, on a 0-100 scale; higher is better.

Metric: Manipulation score (0-100) on placement region (where an object can be placed, with a physical feasibility judgement), 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 November 2028. 76 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3.1 Pro (Preview)55.25#99
2Gemini 3 Pro53.42#145
3Gemini 3 Flash (Preview)50.89#152
4GPT-5.5Non-reasoning48.7#54
5Qwen 3 VL 235B A22B (Thinking)46.49#459
6Claude Sonnet 4.6Thinking43.81#173
7O343.64#237
8Gemini 2.5 Pro42.78#276
9Qwen 3.5 397B A17B42.51#264
10Qwen 3.5 Plus42.46#242
11Doubao-Seed-1.6-vision41.62#361
12Qwen 3.6 Plus41.59#238
13Seed 1.840.79#252
14Qwen 3 VL 30B A3B (Thinking)40.72#719
15Qwen 3 VL 32B (Thinking)40.03#573

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-placement-region · How It Works · Data refreshed daily, snapshot 2026-11-21.