Qwen: Qwen3 235B A22B
● Excellent8.8
/ 10
Qwen3-235B-A22B is a 235B parameter mixture-of-experts (MoE) model developed by Qwen, activating 22B parameters per forward pass. It supports seamless switching between a "thinking" mode for complex reasoning, math, and...
Specifications
| Attribute | Value |
|---|---|
| Lab | Qwen |
| Tags | Open Weight Intelligent Coding Reasoning |
| Overall Score | 8.8/10 |
| Release Date | 2025-04 |
| Context Window | 131,072 tokens |
| Input Price / 1M | $0.45 |
| Output Price / 1M | $1.82 |
| Input Modalities | Text |
| Output Modalities | Text |
Strengths
- Massive 235B MoE architecture (22B active per token)
- Seamless thinking/non-thinking mode switching
- Strong multilingual performance across 100+ languages
- Open-weight with competitive reasoning benchmarks
Weaknesses
- Chinese-first training may bias some cultural references
- Less Western community tooling and ecosystem
- Very large total size limits local deployment
Best For
- Multilingual applications and translation
- Complex reasoning and chain-of-thought
- Cost-efficient open-weight deployment
- Coding and technical tasks
Sources & Further Reading
Related Models
Scores are aggregated from public benchmarks (MMLU, HumanEval, MATH, GSM8K, LMSYS) and normalized to a 1–10 scale. Methodology →