Models / Gemma
Can an RTX 3060 12GB run Gemma 4 E4B?
Small Gemma 4. Google quotes 4.5B effective parameters and 8B once embeddings are included.
- Stored parameters
- 8B
- Active parameters
- Dense
- Q4 total (estimate)
- 5.6 GB
- License
- See model card
Model card. Google's Gemma 4 card: E4B is 4.5B effective parameters and 8B with embeddings. Memory uses 8B because the embedding tables are stored.
Compatibility
| Hardware | Q4 | Q8 | FP16 |
|---|---|---|---|
| Intel Arc Pro B70 Intel · 32 GB VRAM | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run | |
| RTX 5070 Ti NVIDIA · 16 GB VRAM | 9.3 GB Fits estimate No measured run | 16.8 GB Over estimate No measured run | |
| M3 Ultra 512GB Apple · 512 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| 2× NVIDIA GB10 NVIDIA · 256 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M3 Ultra 256GB Apple · 256 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M5 Ultra 256GB Apple · 256 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M2 Ultra 192GB Apple · 192 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| NVIDIA GB10 NVIDIA · 128 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| Ryzen AI Max+ 395 128GB AMD · 128 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M1 Ultra 128GB Apple · 128 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M5 Max 128GB Apple · 128 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| 4× RTX 3090 NVIDIA · 96 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| RTX Pro 6000 Blackwell NVIDIA · 96 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M5 Ultra 96GB Apple · 96 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| Ryzen AI Max+ 395 64GB AMD · 64 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| 2× RTX 3090 NVIDIA · 48 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M4 Pro 48GB Apple · 48 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M4 Max 36GB Apple · 36 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M5 Max 36GB Apple · 36 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| RTX 5090 NVIDIA · 32 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| Radeon AI Pro R9700 AMD · 32 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M4 32GB Apple · 32 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M6 32GB Apple · 32 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| RTX 3090 NVIDIA · 24 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| RTX 4090 NVIDIA · 24 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| TITAN RTX NVIDIA · 24 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| Radeon RX 7900 XTX AMD · 24 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M4 Pro 24GB Apple · 24 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Fits estimate No measured run |
| M3 Pro 18GB Apple · 18 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Tight estimate No measured run |
| RTX 4060 Ti 16GB NVIDIA · 16 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Over estimate No measured run |
| RTX 5060 Ti 16GB NVIDIA · 16 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Over estimate No measured run |
| M4 16GB Apple · 16 GB unified | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Over estimate No measured run |
| RTX 3060 12GB NVIDIA · 12 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Over estimate No measured run |
| RTX 3080 Ti NVIDIA · 12 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Over estimate No measured run |
| RTX 4070 Ti NVIDIA · 12 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Fits estimate No measured run | 16.8 GB Over estimate No measured run |
| RTX 2060 Super NVIDIA · 8 GB VRAM | 5.6 GB Fits estimate No measured run | 9.3 GB Over estimate No measured run | 16.8 GB Over estimate No measured run |
Measured runs
| Hardware | Quant | tok/s | Runtime | Source |
|---|---|---|---|---|
| RTX 5070 Ti | Q4_K_M | 178.6 | llama.cpp b12128 (CUDA) | Source · checked · prefill 10724 |
| Intel Arc Pro B70 | Q4_K_M | 97.3 | llama.cpp b12128 (SYCL) | Source · checked · prefill 4610 |