BLINK
BLINK: Multimodal Large Language Models Can See but Not Perceive. A benchmark for multimodal language models focusing on core visual perception abilities. Reformats 14 classic computer vision tasks into 3,807 multiple-choice questions paired with single or multiple images and visual prompting. Tasks include relative depth estimation, visual correspondence, forensics detection, multi-view reasoning, counting, object localization, and spatial reasoning that humans can solve 'within a blink'.
11rows
scoreprimary metric
2026-05-06sampled
Metadata
Metrics
Score, Normalized Score
| Rank | Subject | Score | Model Match | Provenance | Sampled |
|---|---|---|---|---|---|
| 1 | Qwen3 VL 235B A22B Instruct | 0.71 |  Qwen3 VL 235B A22B Instruct qwen-qwen3-vl-235b-a22b-instruct | Self-reported | 2026-05-06 |
| 2 | Qwen3 VL 8B Instruct | 0.69 | Qwen3 VL 8B Instruct qwen-qwen3-vl-8b-instruct | Self-reported | 2026-05-06 |
| 3 | Qwen3 VL 8B Thinking | 0.69 | Qwen3 VL 8B Thinking qwen-qwen3-vl-8b-thinking | Self-reported | 2026-05-06 |
| 4 | Qwen3 VL 32B Thinking | 0.69 | — | Self-reported | 2026-05-06 |
| 5 | Qwen3 VL 30B A3B Instruct | 0.68 | Qwen3 VL 30B A3B Instruct qwen-qwen3-vl-30b-a3b-instruct | Self-reported | 2026-05-06 |
| 6 | Qwen3 VL 32B Instruct | 0.67 | Qwen3 VL 32B Instruct qwen-qwen3-vl-32b-instruct | Self-reported | 2026-05-06 |
| 7 | Qwen3 VL 235B A22B Thinking | 0.67 | Qwen3 VL 235B A22B Thinking qwen-qwen3-vl-235b-a22b-thinking | Self-reported | 2026-05-06 |
| 8 | Qwen3 VL 4B Instruct | 0.66 | — | Self-reported | 2026-05-06 |
| 9 | Qwen3 VL 30B A3B Thinking | 0.65 | Qwen3 VL 30B A3B Thinking qwen-qwen3-vl-30b-a3b-thinking | Self-reported | 2026-05-06 |
| 10 | Qwen3 VL 4B Thinking | 0.63 | — | Self-reported | 2026-05-06 |
| 11 | Phi-4-multimodal-instruct | 0.61 | — | Self-reported | 2026-05-06 |
No matching rows.