indigoskyai/Qwen2.5-Math-72B-Instruct

TEXT GENERATIONConcurrent Unit Cost:4Model Size:72.7BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 3, 2026License:qwenArchitecture:Transformer Featherless Exclusive Cold

The indigoskyai/Qwen2.5-Math-72B-Instruct is a 72.7 billion parameter instruction-tuned model from the Qwen family, developed by Qwen. It is specifically designed and optimized for solving mathematical problems in both English and Chinese, utilizing Chain-of-Thought (CoT) and Tool-integrated Reasoning (TIR) for enhanced accuracy and complex computations. This model excels in mathematical reasoning tasks, offering significant performance improvements over previous iterations on relevant benchmarks.

Loading preview...

Overview

indigoskyai/Qwen2.5-Math-72B-Instruct is a 72.7 billion parameter instruction-tuned model from the Qwen2.5-Math series, developed by Qwen. This series represents an upgrade from the earlier Qwen2-Math models, focusing on advanced mathematical problem-solving. Unlike its predecessor, Qwen2.5-Math supports both Chain-of-Thought (CoT) and Tool-integrated Reasoning (TIR), enabling it to tackle math problems in both English and Chinese.

Key Capabilities

  • Multilingual Math Solving: Proficient in solving mathematical problems in both English and Chinese.
  • Advanced Reasoning: Leverages CoT for step-by-step reasoning and TIR for precise computation, symbolic manipulation, and algorithmic tasks.
  • Improved Performance: Achieves significant performance gains on Chinese and English mathematics benchmarks using CoT compared to the Qwen2-Math series.
  • High Accuracy: The 72B-Instruct model achieves 87.8 on the MATH benchmark using TIR, demonstrating strong computational accuracy.

Good For

  • Mathematical Problem Solving: Ideal for applications requiring robust solutions to complex math problems.
  • Educational Tools: Can be integrated into platforms for teaching or assisting with mathematical concepts.
  • Research and Development: Suitable for researchers exploring advanced reasoning techniques in LLMs, particularly in the mathematical domain.

Important Note

This model is primarily designed for mathematical tasks. Its performance on other general-purpose tasks is not optimized, and its use for non-mathematical applications is not recommended.

Finetunes
1 models