L4 vs P40: Specs, VRAM & Price
Short answer: choose the L4 when your workload uses FP8 quantized inference or serving. Choose the P40 when your workload is BF16/FP32-only and FP8 buys you nothing.
Both the L4 and the P40 ship with 24 GB of VRAM.
You can rent both by the hour, no purchase required. L4 from $0.649/GPU/hr; P40 from $0.173/GPU/hr. The P40 is 73% cheaper at the entry rate.
L4
Good for inference and light fine-tuning workloads
P40
Good for inference workloads on older-generation hardware
Last updated: 2026-10-10 22:36:10 UTCRefreshes hourly
Specs side by side
| Spec | L4 | P40 |
|---|---|---|
| VRAM | 24 GB | 24 GB |
| Memory bandwidth | 300 GB/s | – |
| Architecture | Ada Lovelace | Pascal |
| Compute capability | 8.9 | 6.1 |
| Precisions in hardware | FP32, FP16, BF16, FP8, INT4 | FP32, FP16 |
| Max board power (TDP) | 72W | – |
| Form factor | – | PCIe |
| Regions | 1 | 1 |
On-demand price
| Rate | L4 | P40 |
|---|---|---|
| Lowest per GPU/hr | $0.649/GPU/hr | $0.173/GPU/hr (cheapest overall) |
Which is better for AI: L4 or P40?
Choose the L4 when your workload uses FP8 quantized inference or serving. Choose the P40 when your workload is BF16/FP32-only and FP8 buys you nothing.
L4 vs P40: common questions
Is the L4 or the P40 better for AI?
Short answer: choose the L4 when your workload uses FP8 quantized inference or serving. Choose the P40 when your workload is BF16/FP32-only and FP8 buys you nothing.
Is the L4 or the P40 cheaper to rent?
On Aquanode's live marketplace the L4 starts at $0.649/GPU/hr (median $0.649/GPU/hr) and the P40 starts at $0.173/GPU/hr (median $0.173/GPU/hr). The P40 is the cheaper of the two at the entry rate, by 73%.
What is the difference between the L4 and the P40?
Both cards report 24 GB of VRAM; the L4 is an Ada Lovelace part and the P40 is Pascal (compute capability 8.9 vs 6.1); only the L4 supports BF16, FP8, INT4 compute. On price, the L4 lists from $0.649/GPU/hr and the P40 from $0.173/GPU/hr.
How widely available are the L4 and the P40?
The L4 is listed in 1 region and the P40 in 1 region right now.
How much does 1,000 GPU-hours cost on the L4 vs the P40?
At the lowest rates listed today, 1,000 GPU-hours costs $649 on the L4 and $173 on the P40, a difference of $476 for the same runtime. Rates are per GPU per hour and update hourly.
How much VRAM does the L4 have compared to the P40?
Both the L4 and the P40 ship with 24 GB of VRAM.
How this comparison is calculated
Every price is normalized to a per-GPU hourly rate using the same pipeline as every other pricing surface on Aquanode, and only the cheapest qualifying offer per provider is shown. A dash means that provider does not currently list that GPU.
Architecture, compute capability and supported precisions come from each vendor's published datasheet for that generation, not from the marketplace feed. This page regenerates at most once per hour.
Memory bandwidth, tensor TFLOPS, NVLink/interconnect and max board power are the same verified-from-datasheet figures shown on each model's own L4 page and P40 page, which link the primary vendor source for that model. A dash here means Aquanode hasn't verified that figure from a primary source yet, never a guess.
Go further
Check what fits in each card's memory with the L4 VRAM calculator and the P40 VRAM calculator, or browse every model by generation in the GPU index.
More L4 comparisons
More P40 comparisons
Related reading: L4 pricing and specs, and The best GPUs for AI, ranked.
