Data center GPU for large-batch inference and professional visualization.
Weekly median price per GPU per hour · Get the data
By provider shows one card per company, placed where its best offer ranked. By configuration lists every offer.
What you can rent comes first: in stock, then waitlist, then not reported, then out of stock. Priced offers rank ahead of quote-only, and on-demand ahead of other billing types.
Within a group, five factors set the order:
We match the provider, its country, GPU model, form factor, billing type, availability and instance name. Matching is partial and case-insensitive. Everyday words work too, so "interruptible" finds spot and "sold out" finds out of stock.
No providers match .
Heads up: A provider's own page may quote a different figure, for example on tax, region or a promotion. A monthly-only plan shows a derived hourly rate. Verify before provisioning. More on how we price.
One A40 has 48 GB of VRAM. In practice, that's enough memory for roughly 60B parameters at 4-bit or 17B at 16-bit, assuming a 32K context. Below are some open-weight LLMs, with the estimated memory and GPUs each one needs.
| Model | Memory (INT4 / FP4) | A40s needed | Cost /hr | Cost /mo |
|---|---|---|---|---|
Qwen3.8-27B
27B
|
19 GB
|
1
|
$0.85
|
$612
|
Gemma 4 31B
30.7B
|
24 GB
|
1
|
$0.85
|
$612
|
GPT-OSS-120B
117B · 5.1B active
|
68 GB
|
2
|
$1.70
|
$1,224
|
GLM-5.3-Flash
320B · 18B active
|
178 GB
|
8
|
$6.80
|
$4,896
|
MiniMax-M3
428B · 23B active
|
241 GB
|
8
|
$6.80
|
$4,896
|
DeepSeek V4.1 Flash
552B · 16B active
|
306 GB
|
8
|
$6.80
|
$4,896
|
Kimi K3
2.8T · 104B active
|
1,545 GB
|
36
|
–
|
–
|
Estimates based on the median on-demand rate. Memory is weights plus FP16/BF16 KV cache at 32K context per request. GPU counts assume 90% of advertised VRAM is usable. No guarantee of runtime support, usable performance, or that a matching quantized build exists. Pricing methodology.
Nvidia A40 · Per GPU
| Compute · dense | |
|---|---|
| FP16 / BF16 | 149.7 TFLOPS 299.4 with sparsity |
| INT8 | 299.3 TOPS 598.6 with sparsity |
| FP32 | 37.4 TFLOPS |
| Precision support | FP16BF16TF32FP32INT4INT8 |
| Memory | |
|---|---|
| Capacity | 48 GB GDDR6 |
| Bandwidth | 696 GB/s |
| Bus width | 384-bit |
| ECC | Yes |
| Silicon | |
|---|---|
| Architecture | Ampere |
| Process | Samsung 8N |
| Transistors | 28.3 billion |
| Shader cores | 10,752 CUDA cores |
| Matrix cores | 336 Tensor cores |
| Compute units | 84 SMs |
| Fabric and host | |
|---|---|
| Host interface | PCIe 4.0 x16 |
| Power | |
|---|---|
| Board power | 300 W |
| Cooling | Passive |
Source: official Nvidia A40 datasheet.
As of September 27, 2026, the median on-demand price is $0.85 per GPU per hour across 6 providers with a priced on-demand config.
| Billing type | Configs | Median /GPU/hr | Cheapest |
|---|---|---|---|
On-demand | 20 | $0.85 | $0.28 (in stock, Vast.ai) |
Reserved | 7 | $0.25 (1 mo, in stock, Vast.ai) |
|
Spot | 4 | $0.31 (in stock, Vast.ai) |
No median for reserved and spot: we only show this when at least 3 providers list the GPU on that billing type.
At 720 hours per month, one A40 costs an estimated $612 at the median on-demand price. Cheapest verified in stock: $202 per month on-demand, $180 reserved (1 mo), $223 spot.
GetDeploying currently tracks A40 configs from 8 providers. The cheapest verified in-stock on-demand configs come from Vast.ai, Runpod, Sesterce and Runcrate. See the full price comparison above for every provider and config.
As of September 27, 2026, the median on-demand price has fallen about 9% over the past 90 days to $0.85 per GPU per hour, though it is about 31% above where it was a year ago.
One A40 runs models up to roughly 60B parameters at 4-bit quantization or 17B at 16-bit, assuming a 32K context. Larger models run across multiple GPUs: the model table above shows the estimated memory and GPU count for popular open-weight LLMs.
48GB GDDR6 with Ampere Tensor Cores. Large VRAM for inference on bigger models. Supports both AI and professional visualization workloads.
No FP8 support. PCIe-only. For pure AI inference, the L40S offers better performance with Ada Lovelace architecture at similar or lower cost.
Nvidia RTX 6000 Ada
48 GB GDDR6 · 38% more bandwidth than the A40
Nvidia L40
48 GB GDDR6 · 24% more bandwidth than the A40
Nvidia A5000
24 GB GDDR6 · NVLink 112 GB/s
Last updated
