OmniCoT: leaderboard

Metric: Overall accuracy (%), question-weighted over the See, Locate and Move dimensions, with chain-of-thought prompting on OmniCoT-B (synthetic 360-degree equirectangular panoramas with three anchor objects given by coordinates and orientation); answers are object identities, visibility, directions or the closest angle option, exact match after normalization. Source: arxiv.org. Saturation forecast: Around 2030. 21 models tracked.

Top models

#ModelScore
1GPT-539.72
2Seed 1.838.9
3Qwen 2.5 VL 72B Instruct35.26
4Grok 434.99
5Qwen 2.5 VL 32B Instruct32.39
6ChatGPT 4o31.58
7GLM-4.5V29.95
8Gemini 3 Flash28.43
9Qwen 3 VL 8B Instruct27.56
10Qwen 2.5 VL 7B Instruct23.53

Interactive version: theaggregate.ai/benchmark?slug=omnicot · How It Works · Data refreshed daily, snapshot 2026-09-29.