NVIDIA RTX A6000 Pricing
Nathanael Chiang3 min read
An NVIDIA RTX A6000 costs $0.33 to $1.09 per GPU-hour to rent on demand from the providers below, against a market median of $0.56 across 17 providers. Beam's A6000 machine is $0.54 an hour with vCPU, RAM, and NVMe included. The A6000 is an Ampere card with 48 GB of ECC memory and NVLink support, and a cheaper route to 48 GB than the L40S.
Key takeaways
- RunPod Community Cloud lists the lowest rate, $0.33, but getdeploying shows it out of stock. Its Secure Cloud is in stock at $0.53.
- Vast.ai's marketplace data shows offers from $0.36 and a median of $0.40, a narrow spread.
- Beam's $0.54 is $0.02 under the median, which getdeploying reports as flat over 90 days.
RTX A6000 cloud prices per hour, by provider
| Provider | SKU | On-demand $/GPU-hr | Notes |
|---|---|---|---|
| RunPod | RTX A6000 (Community Cloud) | $0.33 | Out of stock per getdeploying; Secure Cloud $0.53, in stock |
| Vast.ai | RTX A6000 (marketplace) | $0.36 | Cheapest rentable offer; median offer $0.40; spot from $0.13 per getdeploying; marketplace data from Vast.ai |
| Beam | A6000 PCIe | $0.54 | On-demand machine; vCPU, RAM, and NVMe included |
| Runcrate | RTX A6000 (48GB, PCIe) | $0.63 | Per-second billing; listings run up to $2.13 |
| Lambda | NVIDIA A6000 (1x, 14 vCPUs, 100 GiB RAM) | $1.09 | Same rate for 2x and 4x; out of stock per getdeploying; plus sales tax |
| Market median | — | $0.56 | Across 17 providers with an on-demand price (getdeploying) |
How these prices were collected
Each figure is an on-demand, per-GPU list price for a single-GPU shape where one is offered, with spot rates kept to the notes. Marketplace data from Vast.ai.
- RunPod's pricing page has both cloud tiers.
- Vast.ai's GPU pricing page has the floor and median offers.
- Beam's pricing page has the on-demand machine rate.
- Runcrate's pricing page lists its A6000 rate.
- Lambda's pricing page has the 1x, 2x, and 4x rates.
- getdeploying's RTX A6000 tracker supplies the median.
RTX A6000 vs L40S: the generational trade-off
Both cards have 48 GB, but they're a generation apart. The L40S is an Ada card with FP8 support, which helps inference throughput. The A6000 is Ampere, older and without FP8, but it keeps NVLink for bridging two cards, which the L40S drops.
At market medians, the L40S costs $1.54, 2.75 times the A6000's $0.56. On Beam the gap is $0.22 ($0.76 against $0.54). For fine-tuning, experiments, and jobs where memory matters more than FP8 speed, the A6000 is the cheaper pick. For production inference that can use FP8, price the L40S first.
Where to rent an RTX A6000 for less
Two providers list the A6000 below Beam's $0.54. RunPod Community Cloud is $0.33 when it has stock, and RunPod Secure Cloud is in stock at $0.53, a cent under Beam. Vast.ai's marketplace data shows offers from $0.36, with spot capacity from $0.13 per getdeploying for interruptible jobs. Runcrate is $0.63; Lambda's $1.09 includes 14 vCPUs and 100 GiB of RAM but shows no stock.
Beam's $0.54 includes vCPU, RAM, and NVMe on an on-demand machine. Serverless billing that scales to zero covers Beam's RTX 4090 and RTX 5090, not the A6000.
What this means for your workload
The A6000's 48 GB suits fine-tuning mid-size models, running larger models quantized, and jobs where memory matters more than speed. ECC memory guards long runs against silent memory errors. For NVLink-bridged pairs, confirm the provider offers them. If your model fits in 24 GB, the RTX 4090 costs less: a $0.47 median and $0.44 on Beam.
FAQ
How much does an RTX A6000 cost per hour?
It costs $0.33 to $1.09 per GPU-hour on demand across the providers above, against a $0.56 market median. Beam's A6000 machine is $0.54 an hour.
What's the cheapest way to rent an RTX A6000?
RunPod Community Cloud is cheapest at $0.33 when it has stock, followed by Vast.ai's marketplace from $0.36. If your job can be interrupted, Vast.ai spot capacity starts at $0.13 per getdeploying.
RTX A6000 vs RTX 4090: which should I rent?
Pick the A6000 when you need more than 24 GB in one card. If your model fits in 24 GB, the 4090 costs less at the median ($0.47 vs $0.56) and on Beam ($0.44 vs $0.54), where it can also run serverless.