TAU2
Measured Jul 10, 2026Source
Score
0.27
Models17 1 
Qwen3 235B A22B (Non-reasoning)
阿里巴巴阿里巴巴China
Qwen3 235B A22B is a large-scale Mixture-of-Experts (MoE) language model from Alibaba's Qwen series, with a total of 235 billion parameters but only 22 billion activated per inference. This non-reasoning variant is optimized for general-purpose tasks, offering strong multilingual capabilities, coding proficiency, and efficient performance due to its MoE architecture.
CodingReasoningFastLong contextMultimodal
Input / 1M tokens
$0.70Output / 1M tokens
$2.80Output tokens/s
64.4First-token seconds
1.21sSupported plans
2Benchmark history
Evaluations
Terminalbench Hard
Measured Jul 10, 2026Source
Score
0.06
Lcr
Measured Jul 10, 2026Source
Score
0
Ifbench
Measured Jul 10, 2026Source
Score
0.37
Aime 25
Measured Jul 10, 2026Source
Score
0.24
Aime
Measured Jul 10, 2026Source
Score
0.33
Math 500
Measured Jul 10, 2026Source
Score
0.9
Scicode
Measured Jul 10, 2026Source
Score
0.3
Livecodebench
Measured Jul 10, 2026Source
Score
0.34
Hle
Measured Jul 10, 2026Source
Score
0.05
Gpqa
Measured Jul 10, 2026Source
Score
0.61
Mmlu Pro
Measured Jul 10, 2026Source
Score
0.76
Artificial Analysis Math Index
Measured Jul 10, 2026Source
Score
23.7
Artificial Analysis Intelligence Index
Measured Jul 10, 2026Source
Score
10.9
Tau Banking
Measured Jul 10, 2026Source
Score
0.08
Terminalbench V2 1
Measured Jul 10, 2026Source
Score
0.38
Artificial Analysis Coding Index
Measured Jul 10, 2026Source
Score
36.2
Plan availability
Products and plans that support this model
联通云大模型 Coding Plan
联通云大模型 Coding Plan
China Unicom Cloud's Coding Plan is a subscription service for AI-powered coding, providing access to multiple large language models via an API for use with mainstream AI programming tools. It offers tiered plans with monthly request quotas.
User ratings
Loading ratings...
Discussion
Thinking... Make sure you are connected to GitHub server