Model
Qwen
Qwen is Alibaba's open-weight family, known for excellent multilingual coverage, long context windows, and strong performance on code and reasoning tasks, openly licensed and EU-deployable.
Multilingual, long-context and strong at code
Alibaba
- Excellent multilingual coverage
- Long context windows
- Strong on code and math
- Openly licensed (Apache-2.0 on many models)
Model family
Available variants
Qwen3
Latest generation
Qwen2.5 (7B-72B)
Proven size range
Qwen2.5-Coder
Specialized for code
Qwen2.5-VL
Vision-language
Benchmarks
How it performs
Indicative published results for Qwen2.5 72B Instruct. Scores vary by version, harness and prompt, confirm for the version you deploy.
General knowledge MMLU85
Graduate reasoning GPQA49
Coding HumanEval86
Math MATH83
Model card
What you're deploying, in plain English.
Trained on
Trained on a large multilingual corpus with substantial code and technical content. Varies by version.
Strengths
- Wide language coverage
- Long-context understanding
- Code generation and analysis
Limitations & biases
- Quality varies across smaller variants
- Standard LLM hallucination risk
- Benchmark performance is not a guarantee for your data
Recommended use cases
- Multilingual document processing
- Long-document analysis and RAG
- Developer and data workflows
Prohibited / high-risk
- Unreviewed high-risk classification
- Processing data outside your lawful basis
At a glance
- Vendor
- Alibaba
- Languages
- Very broad multilingual coverage
- License
- Open weights (Apache-2.0 on many models)
- Context
- Long to very long context
- Sizes
- Available in multiple parameter sizes
Hosted in the EU. Runs on vanafter compute or your own infrastructure, your data never leaves European jurisdiction.
Deploy on PryvanBest for: Multilingual teams and long-document or code-heavy workloads.
One arrow. One direction. Forward.
Bring AI into your business, without giving up your data.
Join the waitlist. We're onboarding GDPR-sensitive SMEs across Europe.