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

TEXT GENERATIONPricing:Input $2.72 / Output $4.8Concurrent Unit Cost:2Model Size:32.8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 7, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Qwen2.5-Coder-32B-Instruct is a 32.5 billion parameter instruction-tuned causal language model developed by Qwen, part of the Qwen2.5-Coder series. This model significantly improves code generation, reasoning, and fixing, trained on 5.5 trillion tokens including extensive source code and synthetic data. It supports a full context length of 131,072 tokens and is optimized for real-world code applications, matching the coding abilities of GPT-4o.

Loading preview...

Qwen2.5-Coder-32B-Instruct Overview

Qwen2.5-Coder-32B-Instruct is a 32.5 billion parameter instruction-tuned model from the Qwen2.5-Coder family, developed by Qwen. This series represents a significant advancement over its predecessor, CodeQwen1.5, focusing on enhanced coding capabilities. The model is built on the strong Qwen2.5 foundation and has been trained on an expansive 5.5 trillion tokens, incorporating source code, text-code grounding, and synthetic data.

Key Capabilities

  • Superior Code Performance: Demonstrates significant improvements in code generation, code reasoning, and code fixing, with its coding abilities reportedly matching those of GPT-4o.
  • Comprehensive Foundation for Code Agents: Designed to support real-world applications like Code Agents, maintaining strong performance in mathematics and general competencies alongside its coding strengths.
  • Extended Context Length: Supports an impressive context length of up to 131,072 tokens, with techniques like YaRN for handling even longer texts up to 128K tokens.
  • Robust Architecture: Utilizes a transformer architecture with RoPE, SwiGLU, RMSNorm, and Attention QKV bias, featuring 64 layers and 40 attention heads (GQA).

Good For

  • Advanced Code Generation: Ideal for developers requiring high-quality code generation across various programming tasks.
  • Code Reasoning and Debugging: Suitable for applications that involve understanding, analyzing, and fixing code issues.
  • Building Code Agents: Provides a strong foundation for developing intelligent agents that interact with and manipulate code.
  • Long-Context Code Tasks: Excellent for processing and generating code within very large codebases or complex project contexts due to its extensive context window.

Finetunes
1 models