Alibaba: qwen-math-turbo-latest API

Qwen-math-turbo is the high-efficiency, speed-optimized variant of the Qwen-Math series. It is designed to deliver rapid mathematical reasoning and precise symbolic computation with significantly lower latency than the "Plus" version. By leveraging a specialized distillation process from larger math models, it maintains competitive performance on standard benchmarks like GSM8K and MATH while being optimized for real-time applications. In 2026, the Turbo version features enhanced "Tool-Integrated Reasoning" (TIR), enabling it to invoke a Python interpreter for complex algebraic and statistical operations instantly. It is ideal for cost-sensitive scaling in educational technology and large-scale automated grading.

• Context window: 4,096 tokens
• Max output: 4,096 tokens
• Input: text
• Output: text
• Released: 2025-09-18
• Knowledge cutoff: 2025-10-31