OmniCoT - See: leaderboard
See questions of OmniCoT (2026), a panoramic spatial reasoning benchmark: accuracy in percent on multi-step orientation tracking and relative angle questions over 360-degree images, with chain-of-thought prompting. Higher is better.
Metric: Accuracy (%) on See questions (multi-step orientation tracking, relative angular calculation) 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 May 2027. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 57.29 | #167 |
| 2 | Seed 1.8 | 57.09 | #242 |
| 3 | Grok 4 | 54.4 | #307 |
| 4 | GLM-4.5V | 53.15 | #744 |
| 5 | Qwen 2.5 VL 72B Instruct | 52.03 | #846 |
| 6 | ChatGPT 4o | 51.97 | #610 |
| 7 | Qwen 2.5 VL 32B Instruct | 47.84 | #1053 |
| 8 | Gemini 3 Flash | 47.52 | #164 |
| 9 | Qwen 3 VL 8B Instruct | 41.52 | #981 |
| 10 | Qwen 2.5 VL 7B Instruct | 33.77 | #1833 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=omnicot-see · How It Works · Data refreshed daily, snapshot 2026-11-13.