Llama 3.2 3B
meta-llama/Llama-3.2-3B
Parameters
3.2B
Context
128K
Downloads
478.5K
Likes
887
Architecture
transformer
GPU Requirements
VRAM / RAM / disk estimates by quantization and usage scenario
| Quantization | Minimum | Recommended | Production |
|---|---|---|---|
| FP1616bit | 12.5 GB RAM 32.0 · Disk 48.3 | 28.6 GB RAM 42.9 · Disk 74.3 | 125.8 GB RAM 188.7 · Disk 113.3 |
| FP88bit | 8.7 GB RAM 32.0 · Disk 43.7 | 24.6 GB RAM 37.0 · Disk 69.7 | 121.7 GB RAM 182.6 · Disk 108.7 |
| INT88bit | 8.7 GB RAM 32.0 · Disk 43.7 | 24.6 GB RAM 37.0 · Disk 69.7 | 121.7 GB RAM 182.6 · Disk 108.7 |
| AWQ4bit | 6.9 GB RAM 32.0 · Disk 41.4 | 22.7 GB RAM 34.1 · Disk 67.4 | 119.7 GB RAM 179.6 · Disk 106.4 |
| GPTQ4bit | 6.9 GB RAM 32.0 · Disk 41.4 | 22.8 GB RAM 34.2 · Disk 67.4 | 119.8 GB RAM 179.6 · Disk 106.4 |
| GGUF4bit | 7.0 GB RAM 32.0 · Disk 41.5 | 22.8 GB RAM 34.2 · Disk 67.5 | 119.8 GB RAM 179.7 · Disk 106.5 |
GPU Recommendations
Best ValueBEST
1× RTX 5090
32 GB VRAM
$294/mo
≈ 1156.1 tok/s
Effective production throughput ≈ 809.3 tok/s
score 0.6862
Cheapest
1× L4
24 GB VRAM
$79.5/mo
≈ 193.5 tok/s
Effective production throughput ≈ 135.5 tok/s
score 0.6378
Performance
8× H200 SXM
1128 GB VRAM
$20,440/mo
≈ 9909.7 tok/s
Effective production throughput ≈ 6936.8 tok/s
score 0.2197
Min Complexity
1× RTX 3090
24 GB VRAM
$88.4/mo
≈ 603.9 tok/s
Effective production throughput ≈ 422.7 tok/s
score 0.6422
💬 Community — real deployment experience
0 Articles · 0 Benchmarks
GPUs that fit (single card)
L4
24 GB VRAM
$79.5
/mo · vast
RTX 3090
24 GB VRAM
$88.4
/mo · vast
A10
24 GB VRAM
$176
/mo · vast
RTX 4090
24 GB VRAM
$190
/mo · tensordock
A100 PCIE
80 GB VRAM
$196
/mo · vast