Coding-specialized model family. Code-specific models (0.5B-32B) pre-trained on 5.5T tokens.

Outputs 2

Qwen2.5-Coder

model

Coding-specialized model family. From early 1.5/2.5 versions to the massive Qwen3-Coder-480B (July 2025).

Variants

Name Parameters Notes
Qwen2.5-Coder Initial coding models in 1.5B, 7B, 32B sizes
Qwen3-Coder-480B 480B Released July 2025

Qwen2.5-Coder Technical Report

paper

Code-specific models (0.5B-32B) pre-trained on 5.5T tokens. SOTA on 10+ code benchmarks.

arXiv: 2409.12186

codingopen-weight