Qwen3-235B-A22B at a glance
Qwen3 spans six dense models and two mixture-of-experts models up to 235B parameters, each able to switch between step-by-step thinking and fast answers, with support for 119 languages.
What changed with Qwen3-235B-A22B
On April 29, 2025, Alibaba's Qwen team released Qwen3 with open weights under the Apache 2.0 license. The flagship Qwen3-235B-A22B is a mixture-of-experts model with 235 billion total and 22 billion active parameters; Qwen3-30B-A3B is a smaller MoE, and six dense models range from 0.6B to 32B.
Every Qwen3 model has a thinking mode for step-by-step reasoning and a non-thinking mode for quick replies, which developers can toggle per request. Pretraining used about 36 trillion tokens across 119 languages and dialects, and the team said it strengthened coding, agent behavior and MCP support. The models are on Hugging Face, ModelScope and Kaggle.
Where to compare Qwen3-235B-A22B
Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.
General leaderboards
Why Qwen3-235B-A22B matters for businesses
For businesses, Qwen3 offered a permissively licensed, multilingual family in many sizes, from laptop-scale to server-scale, which made it a popular base for self-hosted assistants and fine-tuning, including for Spanish-language work.
Sources for this story
- Qwen3: Think Deeper, Act Faster (opens in a new tab)Official
Qwen · qwenlm.github.io
- Qwen/Qwen3-235B-A22B (opens in a new tab)Official
Hugging Face · huggingface.co
- Qwen3 235B model page (opens in a new tab)Benchmark
Artificial Analysis · artificialanalysis.ai
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #llm
- #alibaba
- #qwen
- #open-weights
- #multilingual
- #reasoning
