Gemma 3 4B
Parameters
4B
Context Length
131K
Modality
Multimodal
Auxiliary Parameters
430M
Architecture
Dense
KV Cache Rate
60 KB/tok
License
Gemma License
Release Date
12 Mar 2025
Knowledge Cutoff
Aug 2024
API Pricing (per 1M)
Input: $0.05 · Output: $0.10
Hardware Sizing
Precision
Context
Workload Memory: 4.83 GB
INT4 · 8K context
Base Weights: 1.86 GB
KV Cache: 0.47 GB
Runtime & Graphs: 2.50 GB
Minimum Required Hardware
1x NVIDIA L4 24GB
4.8 / 21.6 GB usable
1x RTX 3060 12GB
4.8 / 10.8 GB usable
Apple M4 24GB
4.8 / 18.0 GB usable
Determine requirements for your specific workload:
Open in VRAM Calculator
Architecture Diagram
Evaluation Benchmarks
Rank
#208
| Benchmark | Score | Rank |
|---|---|---|
Professional Knowledge | 0.436 | 60 |
Graduate-Level QA | 0.308 | 120 |
General Text | 1303 | 155 |
Coding Index | 0.03 | 156 |
Intelligence Index | 0.05 | 317 |
Rankings
Overall Rank
#208
Coding Rank
#164
About Gemma 3 4B
Gemma 3 4B is a lightweight vision-language model by Google optimized for efficient execution on consumer GPUs, laptops, and edge devices. It supports multimodal inputs, function calling, and over 140 languages across a 128K context window.
Technical Specifications
Attention
Attention Structure
Grouped-Query Attention
Attention Heads
32
Key-Value Heads
8
Position Embedding
ROPE
Normalization
RMS Normalization
Dimensions
Auxiliary Parameters
430M
Hidden Dimension Size
2,048
Number of Layers
30
About Gemma 3
Gemma 3 is a family of open, lightweight models from Google. It introduces multimodal image and text processing, supports over 140 languages, and features extended context windows up to 128K tokens. Models are available in multiple parameter sizes for diverse applications.