OpenLMM Spatial - BLINK (Circular): leaderboard
BLINK perception tasks on the OpenCompass OpenLMM Spatial Leaderboard (VLMEvalKit), scored with CircularEval (a question counts only if answered correctly under every rotation of its options); values differ from the vanilla accuracy on OpenVLM BLINK. Accuracy in percent, higher is better.
Metric: CircularEval Accuracy (%). Source: huggingface.co. Saturation forecast: Around February 2027. 58 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 61.4 | #159 |
| 2 | GPT-5 | 59.2 | #105 |
| 3 | InternVL3-38B | 58 | #413 |
| 4 | InternVL3-78B | 57.9 | #361 |
| 5 | GPT-4.1 (2025-04-14) | 57.5 | #257 |
| 6 | Gemini 2.5 Flash | 55 | #255 |
| 7 | GPT-5 Mini (2025-08-07) | 54.8 | #175 |
| 8 | InternVL2.5-78B | 54.6 | #434 |
| 9 | Gemini 2.0 Flash | 53.8 | #351 |
| 10 | InternVL3-14B | 53.4 | #506 |
| 11 | GPT-4.1 Mini | 51.2 | #363 |
| 12 | Gemini 1.5 Pro (002) | 48.7 | #368 |
| 13 | InternVL3-8B | 46.1 | #624 |
| 14 | Qwen 2 VL 72B | 44.9 | #450 |
| 15 | Seed 1.6 Flash | 43.5 | #438 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openlmm-spatial-blink-circular · How It Works · Data refreshed daily, snapshot 2026-11-05.