Skip to content

Benchmark

CharXiv (RQ) · Without CI

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

Models measured
10
Scores recorded
12
Sources
1
IQ weight
0.95

Score distribution

82.3%
Mean
84.4%
Median
5.5%
Std dev
77.5%
10th percentile
85.8%
90th percentile
66%
Min
66%88.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.95 for chat models, measured over 10 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 · without Code Interpreter· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.88.4%
2Qwen3.7 PlusRQ · Without Code Interpreter· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.85.8%
3Kimi K3no tools · three-run average · max effort · temp 1.0 · top_p .95· Reported by Moonshot AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.84.8%
4Qwen 3.8 Flash NextQwen3.8-Flash-Next release/model card · without Code Interpreter· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.84.6%
5Qwen3.8 FlashRQ · Without Code Interpreter· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.84.6%
6Qwen3.7 PlusQwen3.7-Plus release · without Code Interpreter· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.84.4%
7Qwen 3.8 27BQwen3.8-27B model card · without Code Interpreter· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.83.7%
8Qwen 3.8 27BRQ · Without Code Interpreter· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.83.7%
9Qwen3.8 Omni FlashRQ · Without Code Interpreter· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.83.5%
10Kimi K2.6Kimi K2.6 official benchmark table · thinking · temp 1.0 · top_p 1.0 · 262,144 context · RQ · no Python · avg@3· Reported by Moonshot AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.80.4%
11Kimi K2.5official model-card base evaluation· Reported by Moonshot AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.77.5%
12Claude Opus 4.6RQ · Without Code Interpreter · max effort· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.66%

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.