Qwen3-235B-A22B
Active Parameters
235B
Context Length
131K
Modality
Text
Architecture
Mixture of Experts (MoE)
License
Apache 2.0
Release Date
29 Apr 2025
Knowledge Cutoff
-
API Pricing (per 1M)
Input: $0.70 · Output: $2.80
System Requirements
VRAM requirements for different quantization methods and context sizes
1,024 tokens
Consumer
24x RTX 4090
24GB VRAM
Datacenter
7x NVIDIA A100
80GB VRAM
Apple Silicon
5x Apple M3 Max
128GB VRAM
131,072 tokens
Consumer
27x RTX 4090
24GB VRAM
Datacenter
8x NVIDIA A100
80GB VRAM
Apple Silicon
6x Apple M3 Max
128GB VRAM
Architecture Diagram
Evaluation Benchmarks
Rank
#139
| Benchmark | Score | Rank |
|---|---|---|
Professional Knowledge | 0.682 | 45 |
Graduate-Level QA | 0.775 | 71 |
General Text | 1403 | 111 |
Agentic Index | 0.01 | 119 |
Coding Index | 0.22 | 138 |
Intelligence Index | 0.08 | 235 |
General Knowledge Reference | 0.878 | 7 |
Coding Archived | 0.60 | 10 |
Rankings
Overall Rank
#139
Coding Rank
#131
About Qwen3-235B-A22B
Qwen3-235B-A22B is Alibaba Cloud's flagship open-source Mixture-of-Experts model activating 22B parameters for advanced computational linguistics and STEM problem solving. It combines dynamic thinking modes with sparse MoE efficiency across 119 languages.
Technical Specifications
Attention
Attention Structure
Grouped-Query Attention
Attention Heads
128
Key-Value Heads
8
Attention Head Dimension
128
Position Embedding
ROPE
RoPE Theta
1,000,000
Sliding Window Attention
No
Sliding Window Size
-
Sliding Window Ratio
-
Linear Attention
-
Linear Attention Ratio
-
Normalization
RMS Normalization
Activation Function
SwigLU
Dimensions
Auxiliary Parameters
-
Hidden Dimension Size
10,240
Number of Layers
100
FFN Intermediate Size (Dense)
1,536
Multi-Token Prediction Heads
-
Tokenizer
Vocabulary Size
151,936
Mixture of Experts
Total Expert Parameters
22.0B
Number of Experts
128
Active Experts
8
Shared Experts
-
FFN Intermediate Size (per Expert)
1,536
Dense Layers Before MoE
-
About Qwen 3
The Alibaba Qwen 3 model family comprises dense and Mixture-of-Experts (MoE) architectures, with parameter counts from 0.6B to 235B. Key innovations include a hybrid reasoning system, offering 'thinking' and 'non-thinking' modes for adaptive processing, and support for extensive context windows, enhancing efficiency and scalability.