L4
efficient inference · video encoding
$200/ 30 days
View GPU L4 vs A40
Compare memory, node configurations, and fixed weekly and monthly cost before choosing between NVIDIA L4 and NVIDIA A40.
A40 provides 24 GB more VRAM per card. A40 costs $8 less per 30-day term . Choose based on whether memory headroom or lower fixed cost is the binding constraint.
efficient inference · video encoding
rendering · inference
| Specification | L4 | A40 |
|---|---|---|
| Architecture | Ada | Ampere |
| VRAM | 24 GB GDDR6 ECC | 48 GB GDDR6 ECC |
| Host resources / card | 8 vCPU · 64 GB RAM | 12 vCPU · 96 GB RAM |
| Local NVMe / card | 1 TB | 1 TB |
| 7-day price / card | $56 | $54 |
| 30-day price / card | $200 | $192 |
| 4-GPU node / 30 days | $736 | $707 |
| Catalog quantity | 315 cards | 265 cards |