Qwen 2.5 Coder 32B Instruct – a code-specialized variant in Alibaba’s Qwen2.5 series.
Matches GPT‑4o on benchmarks and supports over 40 programming languages. Remarkable scores on Aider (~73.7), McEval (~65.9), McEval repair (~75.2)
Qwen: Qwen2.5 Coder 32B Instruct (free)
SOTA among open-source models for code generation, debugging, and reasoning
Context: 132 000 tokens
Output: 8 192 tokens
Modality:
Text
Comparison with other LLMs
Model
Context Window
Multimodal
Latency
Strengths
No items found.
