Qwen2.5-CoderOpen Weight
Released
Qwen2.5-Coder is an AI model released by Qwen on Tuesday, Nov 12 2024, 54 days after Qwen2.5. It is an open-weight model — the trained weights are available to download and run. It was released in 0.5B and 1.5B and 3B and 7B and 14B and 32B parameter sizesThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality..
Get Qwen releases and AI news.
Every other Monday
Available from
Compare Qwen2.5-Coder with
Suggested comparisons
Qwen2.5-CodervsQwen3.8-Max-0902Qwen2.5-CodervsGPT-6.1 TerraQwen2.5-CodervsClaude Haiku 5.5Qwen2.5-CodervsGemini 4 CarbonQwen2.5-CodervsMuse Spark OpenQwen2.5-CodervsGrok 4.7Qwen2.5-CodervsDeepSeek-V4.1-Flash-1017Qwen2.5-CodervsMistral Large 4Qwen2.5-CodervsKimi K3.1 ProQwen2.5-CodervsGLM-5.3-FlashQwen2.5-CodervsNemotron 3.5 Bolt E4B 4GBAbout
Qwen2.5-Coder's November 2024 release completed Alibaba Cloud's coding-model family with six sizes, from 0.5B to 32B parameters. It added 0.5B, 3B, 14B and 32B checkpoints to the earlier 1.5B and 7B models. Each size came in base and instruction-tuned forms, supporting both further training and direct use in programming assistants.
The family targeted code generation, repair and reasoning across programming languages. Deployment details varied by size: the smaller models supported 32K context, while the 7B, 14B and 32B tiers supported longer inputs up to 128K under the documented setup. Most sizes used Apache 2.0, with the 3B model under the Qwen Research License. The release therefore offered several distinct hardware and licensing choices.
Frequently asked questions
Qwen2.5-Coder was released by Qwen on Tuesday, Nov 12 2024.