Qwen3-32B
Parameters
32B
Context Length
131K
Modality
Text
Architecture
Dense
KV Cache Rate
240 KB/tok
License
Apache 2.0
Release Date
29 Apr 2025
Knowledge Cutoff
Aug 2024
API Pricing (per 1M)
Input: $0.08 · Output: $0.28
Hardware Sizing
Precision
Context
Workload Memory: 19.28 GB
INT4 · 8K context
Base Weights: 14.90 GB
KV Cache: 1.88 GB
Runtime & Graphs: 2.50 GB
Minimum Required Hardware
1x NVIDIA L4 24GB
19.3 / 21.6 GB usable
2x RTX 4090 24GB
19.3 / 43.2 GB usable · TP 2
Apple M3 Pro 36GB
19.3 / 27.0 GB usable
Determine requirements for your specific workload:
Open in VRAM Calculator
Architecture Diagram
Evaluation Benchmarks
Rank
#171
| Benchmark | Score | Rank |
|---|---|---|
StackUnseen | 0.457 | 26 |
Agentic Index | auto 0.9 | 118 |
General Text | 1347 | 136 |
Coding Index | 0.15 | 138 |
Intelligence Index | auto 0.07 Standard 0.07 | 270 267 |
Coding Archived | 0.40 | 18 |
Rankings
Overall Rank
#171
Coding Rank
#126
About Qwen3-32B
Qwen3-32B is Alibaba Cloud's premier dense transformer in the Qwen3 series, engineered for autonomous tool orchestration and complex logical deduction. It incorporates hybrid reasoning and Grouped-Query Attention across an extensible 131K context window.
Technical Specifications
Attention
Attention Structure
Grouped-Query Attention
Attention Heads
96
Key-Value Heads
8
Attention Head Dimension
128
Position Embedding
ROPE
RoPE Theta
1,000,000
Normalization
RMS Normalization
Activation Function
SwigLU
Dimensions
Hidden Dimension Size
5,120
Number of Layers
60
FFN Intermediate Size (Dense)
25,600
Tokenizer
Vocabulary Size
151,936
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.