Alibaba logo
Alibaba
Open Weights

Qwen2.5 Coder Instruct 32B

Released Nov 2024

Intelligence
#333
Arena AI
#186
Context131K
Parameters32B

Qwen2.5-Coder-32B-Instruct is an open-weights large language model developed by Alibaba Cloud's Qwen team, specifically optimized for programming and technical tasks. As part of the Qwen2.5-Coder family, this 32-billion parameter model is instruction-tuned to assist with code generation, debugging, and cross-language translation. It was trained on a massive corpus of 5.5 trillion tokens, including high-quality code, mathematics, and textual data. ## Technical Features The model utilizes a dense transformer architecture and supports a context window of 128k tokens, allowing it to process large codebases and complex project files. It demonstrates high proficiency in benchmarks such as HumanEval and MBPP, often performing competitively with significantly larger proprietary models. The model is designed to support various coding-related applications, from simple script writing to complex architectural reasoning.

Rankings & Comparison