Model•active
Qwen3-30B-A3B
Qwen3-30B-A3B is an open-weight Qwen3 mixture-of-experts model with 30B total parameters and 3B active parameters.
Official siteVerified Oct 4, 2026
Overview
The model represents the efficiency-oriented MoE tier of Qwen3, combining a 128K context with multilingual, reasoning, coding and agentic capabilities.
Model specifications
- Family
- Qwen3
- Context
- 128,000 tokens
- Input
- Text
- Output
- Text
- Open weights
- Yes
- Capabilities
- Reasoning, Tool Use, Function Calling