Qwen3 VL 32B Instruct

阿里巴巴China

A 32-billion parameter vision-language model from Alibaba's Qwen3 series, excelling at image understanding, visual question answering, and multimodal reasoning while maintaining good response speed and cost efficiency.

MultimodalReasoningFastCheap

Input / 1M tokens

$0.70

Output / 1M tokens

$2.80

Output tokens/s

54.88

First-token seconds

1.15s

Supported plans

Benchmark history

Evaluations

TAU2

Measured May 14, 2026Source

Score

0.29

Terminalbench Hard

Measured May 14, 2026Source

Score

0.08

Lcr

Measured May 14, 2026Source

Score

0.31

Ifbench

Measured May 14, 2026Source

Score

0.39

Aime 25

Measured May 14, 2026Source

Score

0.68

Scicode

Measured May 14, 2026Source

Score

0.3

Livecodebench

Measured May 14, 2026Source

Score

0.51

Hle

Measured May 14, 2026Source

Score

0.06

Gpqa

Measured May 14, 2026Source

Score

0.67

Mmlu Pro

Measured May 14, 2026Source

Score

0.79

Artificial Analysis Math Index

Measured May 14, 2026Source

Score

68.3

Artificial Analysis Coding Index

Measured May 14, 2026Source

Score

15.6

Artificial Analysis Intelligence Index

Measured May 14, 2026Source

Score

17.2

Aime

Measured May 14, 2026Source

Score

0.94

Math 500

Measured May 14, 2026Source

Score

0.98

Plan availability

Products and plans that support this model

No products or plans have been linked to this model yet.

Discussion

Thinking... Make sure you are connected to GitHub server