Skip to content

Benchmark

TVBench

Vision benchmarks measure how well models understand images. Scores are percentages. Higher is better.

Models measured
6
Scores recorded
7
Sources
2
IQ weight
1.71

Score distribution

76.9%
Mean
78.2%
Median
4%
Std dev
71.5%
10th percentile
80.5%
90th percentile
70.5%
Min
70.5%81.9%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.71 for chat models, measured over 6 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Qwen3.8 MaxQwen3.8-Max release vision/video table · published setting· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.81.9%
2Doubao Seed 2.1 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim80.5%
3Qwen3.7 Plus· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim78.2%
4Qwen3.7 PlusQwen3.7-Plus release · published setting· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.78.2%
5Doubao Seed 2.1 Turbo· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim77.2%
6Doubao Seed 1.8· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim71.5%
7Seed 2.0 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim70.5%

Data credits: Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.