PerceptionTest
multimodal
+
+
+
+
About
Perception Test is a diagnostic multimodal benchmark featuring 11.6k real-world videos that evaluates AI models' perception and reasoning skills across video, audio, and text modalities. This comprehensive assessment tests memory, abstraction, physics understanding, semantics, and various reasoning types including descriptive, explanatory, predictive, and counterfactual reasoning in complex real-world scenarios.
+
+
+
+
Evaluation Stats
Total Models2
Organizations1
Verified Results0
Self-Reported2
+
+
+
+
Benchmark Details
Max Score1
Language
en
+
+
+
+
Performance Overview
Score distribution and top performers
Score Distribution
2 models
Top Score
73.2%
Average Score
71.8%
High Performers (80%+)
0Top Organizations
#1Alibaba Cloud / Qwen Team
2 models
71.8%
+
+
+
+
Leaderboard
2 models ranked by performance on PerceptionTest
| License | Links | ||||
|---|---|---|---|---|---|
| Jan 26, 2025 | tongyi-qianwen | 73.2% | |||
| Jan 26, 2025 | Apache 2.0 | 70.5% |
