Gemma 3 27B
google/gemma-3-27b-it
Google Gemma 3 27B, multimodal chat with 128K context.
Parameters
27B
Context
128K
Downloads
680K
Likes
2.6K
Architecture
transformer
GPU Requirements
VRAM / RAM / disk estimates by quantization and usage scenario
| Quantization | Minimum | Recommended | Production |
|---|---|---|---|
| FP1616bit | 68.6 GB RAM 102.9 · Disk 117.5 | 87.2 GB RAM 130.8 · Disk 143.5 | 187.0 GB RAM 280.5 · Disk 182.5 |
| FP88bit | 36.8 GB RAM 55.1 · Disk 78.2 | 54.0 GB RAM 80.9 · Disk 104.2 | 152.3 GB RAM 228.5 · Disk 143.2 |
| INT88bit | 36.8 GB RAM 55.1 · Disk 78.2 | 54.0 GB RAM 80.9 · Disk 104.2 | 152.3 GB RAM 228.5 · Disk 143.2 |
| AWQ4bit | 21.5 GB RAM 32.2 · Disk 59.4 | 38.0 GB RAM 56.9 · Disk 85.4 | 135.6 GB RAM 203.4 · Disk 124.4 |
| GPTQ4bit | 21.7 GB RAM 32.5 · Disk 59.6 | 38.2 GB RAM 57.2 · Disk 85.6 | 135.8 GB RAM 203.7 · Disk 124.6 |
| GGUF4bit | 22.1 GB RAM 33.1 · Disk 60.1 | 38.6 GB RAM 57.9 · Disk 86.1 | 136.3 GB RAM 204.4 · Disk 125.1 |
GPU Recommendations
1× H100 SXM
80 GB VRAM
≈ 256.9 tok/s
Effective production throughput ≈ 179.8 tok/s
score 0.616
2× RTX 3090
48 GB VRAM
≈ 114.8 tok/s
Effective production throughput ≈ 80.4 tok/s
score 0.6442
8× H200 SXM
1128 GB VRAM
≈ 1177.9 tok/s
Effective production throughput ≈ 824.5 tok/s
score 0.2197
1× RTX A6000
48 GB VRAM
≈ 58.9 tok/s
Effective production throughput ≈ 41.2 tok/s
score 0.6799
💬 Community — real deployment experience
0 Articles · 0 Benchmarks
GPUs that fit (single card)
RTX A6000
48 GB VRAM
$226
/mo · vast
L40S
48 GB VRAM
$341
/mo · vast
RTX PRO 6000 WS
48 GB VRAM
$482
/mo · vast
A100 80GB
80 GB VRAM
$876
/mo · vast
RTX PRO 6000 S
48 GB VRAM
$925
/mo · vast