OpenRouter model
Qwen2.5 Coder 32B Instruct
Qwen2.5-Coder is the latest series of Code-Specific Qwen large language models (formerly known as CodeQwen). Qwen2.5-Coder brings the following improvements upon CodeQwen1.5: - Significantly improvements in **code generation**, **code reasoning**...
- Provider
- OpenRouter
- Availability
- Available
- Model ID
- qwen/qwen-2.5-coder-32b-instruct
- Snapshot
- 2026-10-07
Qwen2.5 Coder 32B Instruct Benchmarks & Agentic Accuracy
Verified primary research & developer preference metrics
What This Means in Real Software Development
- Fits entirely on a single NVIDIA RTX 3090/4090 or Apple M-series Mac (32GB+ unified memory) via 4-bit/8-bit quantization.
- Achieves 84.8% on HumanEval / EvalPlus, matching or exceeding older proprietary models like GPT-4.
- Scores 37.6% on SWE-bench Verified without any cloud API dependency, providing 100% offline air-gapped development.
- Exceptional multi-language code generation across Python, Go, TypeScript, Rust, and C++.
Swarm Multi-Agent Topology
Best Paired With: Claude 3.7 Sonnet (Cloud) or DeepSeek R1 (Local)
The best model for teams with strict data privacy requirements who need autonomous subagent coding completely on-premise.
Integrates natively with Swarm via Ollama or local OpenAI-compatible endpoints (port 11434), maintaining full Git worktree isolation with zero data leaving the host.
What Developers & Practitioners Are Saying
Real-world field observations, production strengths, and developer consensus
"The premier open-weights model for local and self-hosted developer setups. Can run comfortably on a single RTX 4090 / Mac with M-series silicon while matching GPT-4o-level coding."
Verified Real-World Strengths
- Fits on consumer hardware (32B parameters) with high quant fidelity
- Impressive 37.6% on SWE-bench Verified for a model of its size
- Excellent repository understanding and AST manipulation
Production Caveats & Gaps
- Smaller context window than cloud models (128k supported, 32k reliable)
- Can occasionally produce slightly outdated library syntax
Primary Sources & Methodology Citations
- Qwen2.5-Coder Technical Report — Alibaba Cloud Qwen Team (November 2024)
Qwen2.5 Coder 32B Instruct model overview
Qwen2.5 Coder 32B Instruct is listed in the public Swarm catalog through OpenRouter. This page summarizes the verified catalog fields used by Swarm without claiming real-time access for every provider account.
- Provider
- OpenRouter
- Family
- OpenRouter
- Context window
- 32.8K tokens
- Thinking support
- Not supported
- Availability status
- Available
- Catalog ID
- openrouter/qwen/qwen-2.5-coder-32b-instruct
Run Qwen2.5 Coder 32B Instruct in Swarm
Verified CLI workflow & multi-agent architecture
swarm ctl model set --provider openrouter --model openrouter/qwen/qwen-2.5-coder-32b-instruct# Or switch interactively inside the terminal TUI:
/modelsSubagent Role Matrix
Assign Qwen2.5 Coder 32B Instruct to specialized system roles in ~/.swarm/config.json:
- Primary / Planner: Orchestrates full task DAGs and multi-stage pipelines.
- Coder Subagent: Generates code and unit tests inside private Git worktrees.
- Finder Subagent: Performs read-only code search, symbol discovery, and mapping.
Public catalog snapshot. Catalog availability is not a real-time provider status or an access guarantee.