spdev39/Qwen2.5-Coder-32B-Instruct

TEXT GENERATIONConcurrency Cost:2Model Size:32.8BQuant:FP8Ctx Length:32kPublished:Mar 24, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The Qwen2.5-Coder-32B-Instruct is a 32.5 billion parameter instruction-tuned causal language model developed by Qwen, specifically optimized for code generation, reasoning, and fixing. Trained on 5.5 trillion tokens including extensive source code and synthetic data, it offers state-of-the-art coding abilities comparable to GPT-4o. This model also maintains strong performance in mathematics and general competencies, supporting a full context length of 131,072 tokens for complex real-world applications like Code Agents.

Loading preview...

Qwen2.5-Coder-32B-Instruct Overview

Qwen2.5-Coder-32B-Instruct is the instruction-tuned variant of Qwen's latest code-specific large language model series, Qwen2.5-Coder. This 32.5 billion parameter model significantly improves upon its predecessor, CodeQwen1.5, by scaling training tokens to 5.5 trillion, encompassing source code, text-code grounding, and synthetic data. It is designed with a transformer architecture featuring RoPE, SwiGLU, RMSNorm, and Attention QKV bias.

Key Capabilities

  • Advanced Code Generation & Reasoning: Achieves state-of-the-art performance in code generation, reasoning, and fixing, with coding abilities matching GPT-4o.
  • Comprehensive Foundation for Code Agents: Enhances coding capabilities while retaining strong performance in mathematics and general competencies, making it suitable for complex applications like Code Agents.
  • Extended Context Length: Supports a full context length of 131,072 tokens, with a default configuration for 32,768 tokens and YaRN technique for handling even longer texts.

Use Cases

  • Software Development: Ideal for tasks requiring high-quality code generation, debugging, and refactoring across various programming languages.
  • Code Agents: Provides a robust foundation for building intelligent code agents that can understand, generate, and interact with code in complex scenarios.
  • Research & Development: Suitable for exploring advanced applications in AI-assisted programming and large-scale code analysis.