Skip to main content
llm.info

Qwen 2.5 Coder 32B

Open Weights

Alibaba Cloud

Alibaba's state-of-the-art open-source code model achieving GPT-4o-level coding performance. Released November 2024 with 32B parameters - small enough to run on consumer hardware (32GB+ RAM). Best open-source model on multiple code generation benchmarks (EvalPlus, LiveCodeBench, BigCodeBench). Scores 73.7 on Aider code repair (comparable to GPT-4o). Excels across 40+ programming languages with 65.9 on McEval multi-language benchmark. Achieves 75.2 on MdEval code repair (first among open-source). Available in sizes from 0.5B to 32B. Represents breakthrough in accessible, GPT-4o-class coding assistance.

Strengths

  • Matches GPT-4o coding performance as best open-source code model
  • Runs on consumer hardware - 32GB+ RAM sufficient (highly accessible)
  • Best open-source on multiple benchmarks - EvalPlus, LiveCodeBench, BigCodeBench
  • 73.7 Aider score for code repair (comparable to GPT-4o)
  • Exceptional multi-language support - 40+ languages, 65.9 McEval score
  • 75.2 MdEval multi-language repair (first among open-source models)
  • Apache 2.0 license enables unrestricted commercial use

Caveats

  • Text-only (no vision or multimodal capabilities)
  • 4K max output tokens lower than some newer models
  • 32B size still requires significant resources vs smaller models
  • Best performance on code tasks - general knowledge lower than GPT-4o
  • Ecosystem less mature than OpenAI/Anthropic models
Vendor
Alibaba Cloud
Context Window
128,000 tokens
Max Output
4,096 tokens
License
Apache-2.0
Release Date
2024-11-12
Modalities
text
BenchmarkScore
Aider73.7
McEval65.9
MdEval75.2

Capabilities

Vision
Audio
Video
Tool Use

Pricing

Open source (Apache 2.0) - free. Runs on 32GB+ RAM. API providers typically $0.15-$0.60 per 1M input tokens

Reviews

Comments