Skip to content

Benchmark

ScreenSpot

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

Models measured
9
Scores recorded
12
Sources
2
IQ weight
1.69

Score distribution

91.5%
Mean
94.7%
Median
5%
Std dev
83.3%
10th percentile
95.7%
90th percentile
83.3%
Min
83.3%95.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.69 for chat models, measured over 9 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 VL 32B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it.95.8%
2Qwen3 VL 32B Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it.95.7%
3Qwen3 VL 235B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it.95.4%
4Qwen3 VL 235Bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it.95.4%
5qwen3-vl-30b· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it.94.7%
6qwen3-vl-30bthinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it.94.7%
7Qwen3 VL 8B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it.94.4%
8Qwen: Qwen3 VL 8B Thinking· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it.93.6%
9Qwen2.5 VL 72B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim87.1%
10Qwen2.5-VL 7B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim84.7%
11Nova 2 Lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim83.3%
12Nova 2 LiteAmazon Nova 2 Lite official launch chart / technical report · point accuracy, v1· Reported by Amazon Web Services, Amazon AGI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.83.3%

Data credits: 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. · Data by LLM Stats