google/gemma-4-E4B-it
google/gemma-4-E4B-it
Parameters
4.1B
Context
128K
Downloads
5.2M
Likes
1.5K
Architecture
gemma4
42 layers · 2 KV heads · head 256 · hidden 2560
GPU Requirements
VRAM / RAM / disk estimates by quantization and usage scenario
| Quantization | Minimum | Recommended | Production |
|---|---|---|---|
| FP1616bit | 14.0 GB RAM 32.0 · Disk 50.9 | 26.0 GB RAM 38.9 · Disk 76.9 | 94.8 GB RAM 142.2 · Disk 115.9 |
| FP88bit | 9.2 GB RAM 32.0 · Disk 44.9 | 20.9 GB RAM 32.0 · Disk 70.9 | 89.6 GB RAM 134.3 · Disk 109.9 |
| INT88bit | 9.2 GB RAM 32.0 · Disk 44.9 | 20.9 GB RAM 32.0 · Disk 70.9 | 89.6 GB RAM 134.3 · Disk 109.9 |
| AWQ4bit | 6.9 GB RAM 32.0 · Disk 42.1 | 18.5 GB RAM 32.0 · Disk 68.1 | 87.0 GB RAM 130.5 · Disk 107.1 |
| GPTQ4bit | 6.9 GB RAM 32.0 · Disk 42.1 | 18.5 GB RAM 32.0 · Disk 68.1 | 87.1 GB RAM 130.6 · Disk 107.1 |
| GGUF4bit | 7.0 GB RAM 32.0 · Disk 42.2 | 18.6 GB RAM 32.0 · Disk 68.2 | 87.1 GB RAM 130.7 · Disk 107.2 |
GPU Recommendations
1× RTX 5090
32 GB VRAM
≈ 909.6 tok/s
Effective production throughput ≈ 636.7 tok/s
score 0.6521
1× RTX 3090
24 GB VRAM
≈ 475.1 tok/s
Effective production throughput ≈ 332.6 tok/s
score 0.6895
8× H200 SXM
1128 GB VRAM
≈ 7797 tok/s
Effective production throughput ≈ 5457.9 tok/s
score 0.2197
1× RTX 4090D
24 GB VRAM
≈ 511.7 tok/s
Effective production throughput ≈ 358.2 tok/s
score 0.6892
💬 Community — real deployment experience
0 Articles · 0 Benchmarks
GPUs that fit (single card)
RTX 3090
24 GB VRAM
$78.7
/mo · vast
RTX 4090D
24 GB VRAM
$117
/mo · vast
L4
24 GB VRAM
$147
/mo · vast
A10
24 GB VRAM
$176
/mo · vast
RTX 4090
24 GB VRAM
$190
/mo · tensordock