TAU2
Measured May 14, 2026Source
Score
0.53
Models15 1 
Qwen3 235B A22B 2507 (Reasoning)
阿里巴巴阿里巴巴China
This is a reasoning-optimized variant of the Qwen3 235B model from Alibaba Cloud. It is designed to excel in complex logical, mathematical, and coding tasks that require multi-step reasoning. As a large-scale model, it supports long context windows and is part of the advanced Qwen3 series.
ReasoningCodingLong context
Input / 1M tokens
$0.40Output / 1M tokens
$2.15Output tokens/s
59First-token seconds
1.21sSupported plans
2Benchmark history
Evaluations
Terminalbench Hard
Measured May 14, 2026Source
Score
0.14
Lcr
Measured May 14, 2026Source
Score
0.67
Ifbench
Measured May 14, 2026Source
Score
0.51
Aime 25
Measured May 14, 2026Source
Score
0.91
Aime
Measured May 14, 2026Source
Score
0.94
Math 500
Measured May 14, 2026Source
Score
0.98
Scicode
Measured May 14, 2026Source
Score
0.42
Livecodebench
Measured May 14, 2026Source
Score
0.79
Hle
Measured May 14, 2026Source
Score
0.15
Gpqa
Measured May 14, 2026Source
Score
0.79
Mmlu Pro
Measured May 14, 2026Source
Score
0.84
Artificial Analysis Math Index
Measured May 14, 2026Source
Score
91
Artificial Analysis Coding Index
Measured May 14, 2026Source
Score
23.2
Artificial Analysis Intelligence Index
Measured May 14, 2026Source
Score
29.5
Plan availability
Products and plans that support this model
Synthetic
Synthetic
Synthetic is a platform that runs open-source AI models in private, secure datacenters. It offers subscription and usage-based pricing for accessing models like DeepSeek, Llama, MiniMax, Kimi, and others, with OpenAI-compatible API access for use in coding agents and tools.
Discussion
Thinking... Make sure you are connected to GitHub server