Skip to content

Benchmark

QwenSVGBench

Coding benchmarks compare models on programming tasks. Scores are Elo ratings. Higher ratings are better.

Models measured
6
Scores recorded
7
Sources
1
IQ weight
0.68

Score distribution

1,643.4
Mean
1,648
Median
80.5
Std dev
1,588
10th percentile
1,713
90th percentile
1,499
Min
1,4991,758
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.68 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
1GPT-5.6 SolQwen内部双语 SVG 基准 · 自动渲染 + 多模态评判 · BT/Elo· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,758
2Qwen3.8 MaxQwen内部双语 SVG 基准 · 自动渲染 + 多模态评判 · BT/Elo· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,713
3Claude Fable 5Qwen内部双语 SVG 基准 · 自动渲染 + 多模态评判 · BT/Elo· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,690
4Claude Opus 4.8Qwen内部双语 SVG 基准 · 自动渲染 + 多模态评判 · BT/Elo· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,648
5Qwen3.7 MaxQwen3.7-Max release · QwenSVG · BT/Elo· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,608
6Qwen3.7 PlusQwen3.7-Plus release · QwenSVG · BT/Elo· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,588
7Qwen3.7 MaxQwen内部双语 SVG 基准 · 自动渲染 + 多模态评判 · BT/Elo· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,499

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.