Paper 2603.00729
Qwen3-Coder-Next Technical Report
01 In brief
Summary
Qwen3-Coder-Next is an 80-billion-parameter open-weight language model with only 3 billion active parameters per forward pass, designed for coding agents.
It is built on Qwen3-Next with hybrid attention and Mixture-of-Experts, and trained via a staged pipeline: continued pretraining on code and agent-centric data, supervised fine-tuning, and reinforcement learning.
The model uses large-scale synthesis of verifiable, executable tasks (e.g., from GitHub PRs and bug injection) and an internal orchestration system (MegaFlow) for scalable training.
It achieves competitive results on SWE-Bench Verified (70.6% with SWE-Agent), SWE-Bench Multilingual (62.8%), SWE-Bench Pro (42.7%), and Terminal-Bench 2.0 (36.2% with Terminus-2 json), matching or outperforming larger models like DeepSeek-V3.2 and GLM-4.7.
The model also shows strong performance on general coding benchmarks (e.g., LiveCodeBench 58.93%, Codeforces 2100) and competitive math (AIME25 83.07%), while maintaining general knowledge.
Training included diverse tool-call templates to improve format generalization, and expert distillation from web development, UX, single-turn RL, and software engineering experts.
Limitations include gaps in complex large-scale software engineering tasks and frontend/UI capabilities.
02 From the paper
Abstract
We present Qwen3-Coder-Next, an open-weight language model specialized for coding agents. Qwen3-Coder-Next is an 80-billion-parameter model that activates only 3 billion parameters during inference, enabling strong coding capability with efficient inference. In this work, we explore how far strong training recipes can push the capability limits of models with small parameter footprints. To achieve this, we perform agentic training through large-scale synthesis of verifiable coding tasks paired with executable environments, allowing learning directly from environment feedback via mid-training and reinforcement learning. Across agent-centric benchmarks including SWE-Bench and Terminal-Bench, Qwen3-Coder-Next achieves competitive performance relative to its active parameter count. We release both base and instruction-tuned open-weight versions to support research and real-world coding agent development.