L40 vs RTX A6000: Specs, VRAM & Price
Short answer: choose the L40 when your workload uses FP8 quantized inference or serving. Choose the RTX A6000 when your workload is BF16/FP32-only and FP8 buys you nothing.
Both the L40 and the RTX A6000 ship with 48 GB of VRAM.
You can rent both by the hour, no purchase required. L40 from $0.850/GPU/hr; RTX A6000 from $0.550/GPU/hr. The RTX A6000 is 35% cheaper at the entry rate.
L40
Good for fine-tuning mid-size models and high-throughput inference
RTX A6000
Good for training and inference on models that don't need FP8 kernels
Last updated: 2026-11-03 UTCRefreshes every 6 hours
Specs side by side
| Spec | L40 | RTX A6000 |
|---|---|---|
| VRAM | 48 GB | 48 GB |
| Memory bandwidth | 864 GB/s | 768 GB/s |
| Architecture | Ada Lovelace | Ampere |
| Compute capability | 8.9 | 8.6 |
| Precisions in hardware | FP32, FP16, BF16, FP8, INT4 | FP32, FP16, BF16, INT4 |
| Peak FP16/BF16 tensor TFLOPS | 181.05 TFLOPS | – |
| Peak FP8 tensor TFLOPS | 362 TFLOPS | – |
| NVLink / multi-GPU interconnect | – | NVLink, 2-way |
| Max board power (TDP) | 300W | 300W |
| Regions | 2 | 3 |
On-demand price
| Rate | L40 | RTX A6000 |
|---|---|---|
| Lowest per GPU/hr | $0.850/GPU/hr | $0.550/GPU/hr |
Which is better for AI: L40 or RTX A6000?
Choose the L40 when your workload uses FP8 quantized inference or serving. Choose the RTX A6000 when your workload is BF16/FP32-only and FP8 buys you nothing.
L40 vs RTX A6000: common questions
Is the L40 or the RTX A6000 better for AI?
Short answer: choose the L40 when your workload uses FP8 quantized inference or serving. Choose the RTX A6000 when your workload is BF16/FP32-only and FP8 buys you nothing.
Is the L40 or the RTX A6000 cheaper to rent?
On Aquanode's live marketplace the L40 starts at $0.850/GPU/hr (median $1.03/GPU/hr) and the RTX A6000 starts at $0.550/GPU/hr (median $0.627/GPU/hr). The RTX A6000 is the cheaper of the two at the entry rate, by 35%.
What is the difference between the L40 and the RTX A6000?
Both cards report 48 GB of VRAM; the L40 is an Ada Lovelace part and the RTX A6000 is Ampere (compute capability 8.9 vs 8.6); only the L40 supports FP8 compute. On price, the L40 lists from $0.850/GPU/hr and the RTX A6000 from $0.550/GPU/hr.
How widely available are the L40 and the RTX A6000?
The L40 is listed in 2 regions and the RTX A6000 in 3 regions right now.
How much does 1,000 GPU-hours cost on the L40 vs the RTX A6000?
At the lowest rates listed today, 1,000 GPU-hours costs $850 on the L40 and $550 on the RTX A6000, a difference of $300 for the same runtime. Rates are per GPU per hour and update hourly.
How much VRAM does the L40 have compared to the RTX A6000?
Both the L40 and the RTX A6000 ship with 48 GB of VRAM.
Does the L40 or the RTX A6000 support NVLink?
The RTX A6000 supports NVLink, 2-way; the L40 has no dedicated GPU-to-GPU interconnect.
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 L40 page and RTX A6000 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 L40 VRAM calculator and the RTX A6000 VRAM calculator, see vendor-published throughput in the GPU benchmarks and specs table, or browse every model by generation in the GPU index.
More L40 comparisons
More RTX A6000 comparisons
Related reading: L40 pricing and specs, The NVIDIA Inception program, explained, Google Colab alternatives for dedicated GPU access, Free GPU credits for students and researchers, and RunPod volume disk vs. network volume, compared.
