Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Qwen3-Coder-Next Technical Report

Ruisheng Cao, Mouxiang Chen, Jiawei Chen, Zeyu Cui, Yunlong Feng, Binyuan Hui, Yuheng Jing, Kaixin Li, Mingze Li, Junyang Lin, Zeyao Ma, Kashun Shum, Xuwu Wang, Jinxi Wei, Jiaxi Yang, Jiajun Zhang, Lei Zhang, Zongmeng Zhang, Wenting Zhao, Fan Zhou

TL;DR

This work explores how far strong training recipes can push the capability limits of models with small parameter footprints and presents Qwen3-Coder-Next, an open-weight language model specialized for coding agents that activates only 3 billion parameters during inference, enabling strong coding capability with efficient inference.

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.

Qwen3-Coder-Next Technical Report

TL;DR

This work explores how far strong training recipes can push the capability limits of models with small parameter footprints and presents Qwen3-Coder-Next, an open-weight language model specialized for coding agents that activates only 3 billion parameters during inference, enabling strong coding capability with efficient inference.

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.
Paper Structure (61 sections, 2 equations, 8 figures, 16 tables)

This paper contains 61 sections, 2 equations, 8 figures, 16 tables.

Table of Contents

  1. Introduction
  2. Scaling up Agentic Training
  3. Task Synthesis
  4. Creating Executable Environments from GitHub PRs.
  5. Synthesizing Issues.
  6. Infrastructure
  7. Mid-training
  8. Data
  9. Natural Data
  10. GitHub.
  11. Text–code Grounding Data.
  12. GitHub Pull Requests (PRs).
  13. Synthetic Data
  14. Single-turn QA.
  15. Multi-turn Agentic Coding.
  16. ...and 46 more sections

Figures (8)

  • Figure 1: Comparison of Qwen3-Coder-Next to other open-weight models on SWE-Bench Verified, SWE-Bench Multilingual, SWE-Bench Pro, Terminal-Bench 2.0, and Aider.
  • Figure 2: Our pipeline for synthesizing bugs to scale up the number of software engineering tasks.
  • Figure 3: Scaling analysis of code agent pretraining on SWE tasks, SWE-Bench Verified and SWE-Bench Multilingual. Left: Within-scaffold scaling shows consistent improvement with the increase of mid-training tokens. Right: Cross-scaffold transfer reveals limited generalization across different agent frameworks.
  • Figure 4: Different tool chat templates, split by tool definition, tool calls, and tool response.
  • Figure 5: SWE-bench Verified performance vs. number of tool chat templates. Data volume and training configuration are kept identical.
  • ...and 3 more figures