CodeQwen1.5-7B
Open 7B code model pretrained on about 3T tokens of code, covering 92 programming languages with 64K context.
A coding specialist on Qwen1.5, positioned against StarCoder2, CodeLlama and DeepSeek-Coder as an open alternative to proprietary assistants. Precursor to Qwen2.5-Coder and Qwen3-Coder.
- Date
- Tuesday, 16 April 2024
- Lab
- Alibaba (Qwen)
- Kind
- open-weights
- Access
- open weights (restricted license)
Figures
| Measure | Value | Measured by |
|---|---|---|
| Code pretraining tokens | about 3T 92 programming languages; 64K context | company |
Blog dated 2024-04-16. License not checked; benchmark claims not reproduced.
Sources
- qwenlm.github.io/blog/codeqwen1.5/
- huggingface.co/api/models?author=Qwen&sort=createdAt&direction=1&limit=100
This record was checked against its sources on 6 October 2026. How we check