Alibaba Cloud·Qwen

Qwen3-235B-A22B

The flagship Qwen3 MoE, combining thinking and non-thinking modes in one open model.

CurrentLanguageReasoningCodingAgentOpen weights

MODEL NUMBER022

Specifications

Only published specifications are shown.

Total parameters

235B

Active parameters

22B

Architecture

MoE

Context

128K

Input modalities

Text

Open weights

Yes

License

Apache-2.0

Status

Current

The flagship Qwen3 MoE, combining thinking and non-thinking modes in one open model.

Qwen3 brought hybrid reasoning and efficient MoE together in an Apache 2.0 open-weight model.

Compared with Qwen2-72B ↗

PreviousQwen2-72B72.7B · 128K ctx

CurrentQwen3-235B-A22B235B · 128K ctx

New capabilities Hybrid reasoning Tool use

GQA

Hybrid reasoningCodingTool useMultilingual

2025-04-29Open weights

Qwen3 launched with the 235B-A22B MoE as flagship.

Select a card to view the related model.