Gemma 3 4B
by Google · gemma-3 family
4B
parameters
text-generation code-generation reasoning multilingual vision summarization
Gemma 3 4B is the smallest multimodal model in the Gemma 3 family, supporting both text and image inputs. It delivers impressive performance for its size, outperforming many larger text-only models on standard benchmarks. With 128K context and vision support at just 5 GB VRAM (Q4), it's an excellent choice for users with 8 GB GPUs who want multimodal capabilities. Drag images into Ollama's desktop app to ask questions about them.
Quick Start with Ollama
ollama run 4b-it-q4_K_M | Creator | |
| Parameters | 4B |
| Architecture | transformer-decoder |
| Context | 128K tokens |
| Released | Mar 12, 2025 |
| License | Gemma Terms of Use |
| Ollama | gemma3:4b |
Quantization Options
| Format | File Size | VRAM Required | Quality | Ollama Tag |
|---|---|---|---|---|
| Q4_K_M rec | 3.3 GB | 5 GB |
★
★
★
★
★
| 4b-it-q4_K_M |
| Q8_0 | 5 GB | 7.5 GB |
★
★
★
★
★
| 4b-it-q8_0 |
| F16 | 8.6 GB | 11.5 GB |
★
★
★
★
★
| 4b-it-fp16 |
Compatible Hardware
for Q4_K_M (5 GB VRAM)
Compatible Hardware
| Hardware | VRAM | Type | Fit | Est. Speed |
|---|---|---|---|---|
| Mac Studio M4 Ultra 512GB | 512 GB | mac | Runs | ~164 tok/s |
| Mac Pro M2 Ultra 192GB | 192 GB | mac | Runs | ~160 tok/s |
| Mac Studio M4 Ultra 192GB | 192 GB | mac | Runs | ~164 tok/s |
| Mac Studio M4 Max 128GB | 128 GB | mac | Runs | ~109 tok/s |
| MacBook Pro M4 Max 128GB | 128 GB | mac | Runs | ~109 tok/s |
| MacBook Pro M3 Max 96GB | 96 GB | mac | Runs | ~80 tok/s |
| Mac mini M4 Pro 64GB | 64 GB | mac | Runs | ~55 tok/s |
| Mac Studio M4 Max 64GB | 64 GB | mac | Runs | ~109 tok/s |
| MacBook Pro M4 Max 64GB | 64 GB | mac | Runs | ~109 tok/s |
| Mac mini M4 Pro 48GB | 48 GB | mac | Runs | ~55 tok/s |
| MacBook Pro M3 Max 48GB | 48 GB | mac | Runs | ~80 tok/s |
| MacBook Pro M4 Max 48GB | 48 GB | mac | Runs | ~109 tok/s |
| MacBook Pro M4 Pro 48GB | 48 GB | mac | Runs | ~55 tok/s |
| Mac Studio M4 Max 36GB | 36 GB | mac | Runs | ~109 tok/s |
| MacBook Pro M3 Pro 36GB | 36 GB | mac | Runs | ~30 tok/s |
| NVIDIA GeForce RTX 5090 | 32 GB | gpu | Runs | ~358 tok/s |
| iMac M4 32GB | 32 GB | mac | Runs | ~24 tok/s |
| Mac mini M4 32GB | 32 GB | mac | Runs | ~24 tok/s |
| MacBook Air M4 32GB | 32 GB | mac | Runs | ~24 tok/s |
| AMD Radeon RX 7900 XTX | 24 GB | gpu | Runs | ~192 tok/s |
| NVIDIA GeForce RTX 3090 | 24 GB | gpu | Runs | ~187 tok/s |
| NVIDIA GeForce RTX 4090 | 24 GB | gpu | Runs | ~202 tok/s |
| iMac M3 24GB | 24 GB | mac | Runs | ~20 tok/s |
| Mac mini M2 24GB | 24 GB | mac | Runs | ~20 tok/s |
| Mac mini M4 Pro 24GB | 24 GB | mac | Runs | ~55 tok/s |
| MacBook Air M2 24GB | 24 GB | mac | Runs | ~20 tok/s |
| MacBook Air M4 24GB | 24 GB | mac | Runs | ~24 tok/s |
| MacBook Pro M4 Pro 24GB | 24 GB | mac | Runs | ~55 tok/s |
| AMD Radeon RX 7900 XT | 20 GB | gpu | Runs | ~160 tok/s |
| MacBook Pro M3 Pro 18GB | 18 GB | mac | Runs | ~30 tok/s |
| AMD Radeon RX 6800 XT | 16 GB | gpu | Runs | ~102 tok/s |
| AMD Radeon RX 7800 XT | 16 GB | gpu | Runs | ~125 tok/s |
| Intel Arc A770 | 16 GB | gpu | Runs | ~112 tok/s |
| NVIDIA GeForce RTX 4060 Ti 16GB | 16 GB | gpu | Runs | ~58 tok/s |
| NVIDIA GeForce RTX 4070 Ti Super | 16 GB | gpu | Runs | ~134 tok/s |
| NVIDIA GeForce RTX 4080 Super | 16 GB | gpu | Runs | ~147 tok/s |
| NVIDIA GeForce RTX 4080 | 16 GB | gpu | Runs | ~143 tok/s |
| NVIDIA GeForce RTX 5070 Ti | 16 GB | gpu | Runs | ~179 tok/s |
| NVIDIA GeForce RTX 5080 | 16 GB | gpu | Runs | ~192 tok/s |
| iMac M1 16GB | 16 GB | mac | Runs | ~14 tok/s |
| iMac M4 16GB | 16 GB | mac | Runs | ~24 tok/s |
| Mac mini M1 16GB | 16 GB | mac | Runs | ~14 tok/s |
| Mac mini M4 16GB | 16 GB | mac | Runs | ~24 tok/s |
| MacBook Air M2 16GB | 16 GB | mac | Runs | ~20 tok/s |
| MacBook Air M3 16GB | 16 GB | mac | Runs | ~20 tok/s |
| MacBook Air M4 16GB | 16 GB | mac | Runs | ~24 tok/s |
| MacBook Pro M1 16GB | 16 GB | mac | Runs | ~14 tok/s |
| MacBook Pro M2 Pro 16GB | 16 GB | mac | Runs | ~40 tok/s |
| AMD Radeon RX 7700 XT | 12 GB | gpu | Runs | ~86 tok/s |
| NVIDIA GeForce RTX 3060 12GB | 12 GB | gpu | Runs | ~72 tok/s |
| NVIDIA GeForce RTX 3080 12GB | 12 GB | gpu | Runs | ~182 tok/s |
| NVIDIA GeForce RTX 4070 Super | 12 GB | gpu | Runs | ~101 tok/s |
| NVIDIA GeForce RTX 4070 Ti | 12 GB | gpu | Runs | ~101 tok/s |
| NVIDIA GeForce RTX 4070 | 12 GB | gpu | Runs | ~101 tok/s |
| NVIDIA GeForce RTX 5070 | 12 GB | gpu | Runs | ~134 tok/s |
| NVIDIA GeForce RTX 2080 Ti | 11 GB | gpu | Runs | ~123 tok/s |
| NVIDIA GeForce RTX 3080 10GB | 10 GB | gpu | Runs | ~152 tok/s |
| AMD Radeon RX 7600 | 8 GB | gpu | Runs | ~58 tok/s |
| Intel Arc A750 | 8 GB | gpu | Runs | ~102 tok/s |
| NVIDIA GeForce RTX 3060 Ti | 8 GB | gpu | Runs | ~90 tok/s |
| NVIDIA GeForce RTX 3070 | 8 GB | gpu | Runs | ~90 tok/s |
| NVIDIA GeForce RTX 4060 Ti 8GB | 8 GB | gpu | Runs | ~58 tok/s |
| NVIDIA GeForce RTX 4060 | 8 GB | gpu | Runs | ~54 tok/s |
| MacBook Air M1 8GB | 8 GB | mac | Runs | ~14 tok/s |
| MacBook Air M2 8GB | 8 GB | mac | Runs | ~20 tok/s |
Benchmark Scores
62.0
mmlu