Qwen Models
Chinese open-source • Alibaba research
6 Models
Latest: Qwen3-Max
Qwen3-Max
Released September 2025
Alibaba's flagship model with over 1 trillion parameters and exceptional reasoning
Parameters
1T+
Context
262,144 tokens
Key Features
Massive Scale Long Context Multilingual
+1 more
View Details →
Qwen3-235B-A22B
Released April 2025
Alibaba's flagship open-source MoE model with 235B total parameters and 22B active
Parameters
235B (22B active)
Context
128,000 tokens
Key Features
Open Source MoE Architecture Efficient Inference
+1 more
View Details →
QwQ-32B
Released March 2025
Reasoning-focused model with extended thinking capabilities
Parameters
32 billion
Context
32,768 tokens
Key Features
Extended Thinking Chain-of-Thought Math Reasoning
+1 more
View Details →
Qwen3-30B-A3B
Released April 2025
Efficient MoE model with 30B total but only 3B active parameters
Parameters
30B (3B active)
Context
128,000 tokens
Key Features
MoE Architecture Efficient Inference Cost Effective
+1 more
View Details →
Qwen2.5-VL-32B
Released January 2025
Vision-language model with strong multimodal understanding
Parameters
32 billion
Context
32,768 tokens
Key Features
Vision-Language Image Understanding OCR
+1 more
View Details →
Qwen2.5-72B-Instruct
Released September 2024
Alibaba's instruction-tuned flagship open-source model
Parameters
72 billion
Context
128,000 tokens
Key Features
Open Source Instruction Following Long Context
+1 more
View Details →
