← Back to Docs

Qwen 2.5 Coder 32B

Alibaba's code-specialized model with strong implementation and debugging capabilities.

ProviderAlibaba Cloud (Qwen team)
Parameters32 billion
Context128K tokens
Best forCode generation, debugging, refactoring
Dispatch roleimplement_agent, frontend_agent
InfrastructureDispatch Pro cloud
CostIncluded in Pro ($20/month, shared 5M token pool)

Overview

Qwen 2.5 Coder 32B is purpose-built for code tasks. It handles multi-file edits, understands project structure, and generates idiomatic code across TypeScript, Python, Rust, Go, and other languages. In Dispatch, it serves as a cloud implementation agent when local agents are unavailable or when you want to parallelize work.

When Dispatch uses Qwen Coder

  • Implementation tasks -- Building features, fixing bugs, writing functions from specs.
  • Frontend tasks -- React components, CSS, responsive layouts.
  • Code review -- Post-build quality analysis alongside local agents.

Configuration

Available to all Dispatch Pro subscribers. Set any pipeline stage to qwen-coder-32b in Settings > Pipeline Profiles, or let Auto Model route automatically.

Isolation

All cloud model runs use worktree isolation. Changes are reviewed through the standard HITL approval flow before merging.