Alibaba releases Qwen3-Coder-Next to rival OpenAI, Anthropic
marktechpost.com
February 4, 2026
5 min read
🔥🔥🔥🔥🔥
39/100
Summary
Qwen3-Coder-Next is an open-weight language model designed for coding agents and local development, built on the Qwen3-Next-80B-A3B backbone. It features a sparse Mixture-of-Experts (MoE) architecture with 80 billion total parameters, activating only 3 billion parameters per token to optimize performance and reduce inference costs.
Key Takeaways
- Qwen3-Coder-Next is an open-weight language model specifically designed for coding agents and local development, built on the Qwen3-Next-80B-A3B backbone.
- The model employs a sparse Mixture-of-Experts architecture with 80 billion total parameters, activating only 3 billion parameters per token to optimize performance and reduce inference costs.
- Qwen3-Coder-Next is trained on approximately 800,000 executable tasks using reinforcement learning, enabling it to perform long-horizon reasoning and tool sequencing.
- The model achieves competitive benchmark scores, such as 70.6 on SWE-Bench, indicating performance comparable to larger models with significantly more active parameters.
![[AINews] Why OpenAI Should Build Slack](https://substackcdn.com/image/fetch/$s_!XQAE!,w_1200,h_675,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F89ee056a-0ea2-4473-8e1c-9b21f034c717_1474x2116.png)