# GPUBank vs Novita AI

> Novita AI is an AI cloud that leads with serverless model APIs and agent sandboxes, and also sells per-second GPU container instances, serverless GPU endpoints and 8-GPU bare metal nodes; its terms name no legal entity beyond Novita AI. Its H100 SXM costs $3.39 per GPU-hour on demand and $1.70 on spot; GPUBank charges $1.85 on demand. Novita has serverless GPUs, SOC 2 Type II and PayPal; GPUBank has a wider on-demand catalogue, crypto top-ups and named regions.

- Page: https://gpubank.io/compare/gpubank-vs-novita
- Prices last checked on 29 September 2026, on Novita AI's own pages (sources below). Page updated 2 November 2026.
- Novita AI: https://novita.ai
- GPUBank: https://gpubank.io, GPU cloud billed per second at an all-in hourly price (vCPU, RAM, NVMe, one IPv4 and traffic included), prepaid balance topped up in crypto.

## Verdict

**Serverless GPUs and SOC 2 at Novita AI. Lower on-demand prices and crypto top-ups at GPUBank.**

Novita suits teams that already use its model APIs and want GPU containers, serverless endpoints and spot capacity on the same balance. Its public GPU list has five instance models, and for the GPUs both of us rent on demand GPUBank costs up to 45 % less, with the L40S at the same $0.55. Novita's own pages disagree on some prices, for example the RTX 5090 at $0.56 or $0.73 on demand. Pay-as-you-go instances release their hardware when stopped and may not restart, and stopped data is deleted after 7 days.

Score over 24 points: GPUBank ahead on 10, even on 8, Novita AI ahead on 6.

## Key facts

| Criterion | Novita AI | GPUBank |
|---|---|---|
| H100 SXM, per GPU-hour | $3.39 on demand, $1.70 spot | $1.85, all-in |
| Cheapest GPU-hour | $0.33, RTX 4090 | $0.23, RTX 4090 |
| Billing granularity | Per second, settled hourly | Per second |
| Stopped instance | Disk kept 7 days at $0.005/GB-day | Disk only |
| Persistent storage | Network volume $0.06/GB-month | $0.03 per GB-month |
| Egress traffic | Included | Included |
| Crypto payments | Not supported | Bitcoin, USDT, Ethereum and 12+ more |
| Card payments | Cards via Stripe, PayPal | No |
| Minimum deposit | More than $10 | $25 |
| GPU models listed | 7 | 10 |
| Regions | Not published | 4 |
| Serverless GPUs | Serverless GPU endpoints | No |
| Availability target | No figure; contractual SLAs on bare metal | 99.9 % monthly |
| Compliance reports | SOC 2 Type II | None published |
| Support | Email, Discord, sales; 24/7 claimed | Tickets from a $25 balance |
| Headquarters | Not published | Tallinn, Estonia |

## Same GPUs, price per GPU-hour (USD, excl. VAT)

| GPU | Novita AI | GPUBank | Difference |
|---|---|---|---|
| B200 SXM | $4.77 (reserved) | $4.15 | -13 % |
| H100 SXM | $3.39 (on demand) | $1.85 | -45 % |
| L40S | $0.55 (on demand) | $0.55 | 0 % |
| RTX 5090 | $0.56 (on demand) | $0.48 | -14 % |
| RTX 4090 | $0.33 (on demand) | $0.23 | -30 % |

## When to pick Novita AI

- Serverless GPU endpoints with autoscaling and per-second worker billing
- 8x H100 SXM bare metal at $1.70 per GPU-hour, with NVLink and RDMA
- SOC 2 Type II, report under NDA, and a DPA with SCCs
- Pay by card or PayPal; RTX 6000 Ada, which GPUBank does not list

## When to pick GPUBank

- H100 SXM at $1.85 on demand, against $3.39 at Novita
- B200, H200, H100 PCIe, A100, RTX PRO 6000 and MI300X on demand
- Stop an instance and keep its disk; Novita deletes it after 7 days
- Pay in Bitcoin, USDT and 12+ other coins; Novita takes no crypto
- Public IPv4 included and four named regions

## Questions

### Is GPUBank cheaper than Novita AI for an H100?

On demand, yes. Novita lists the H100 SXM at $3.39 per GPU-hour on demand and $1.70 on spot. GPUBank charges $1.85 on demand and $1.11 on interruptible instances, with vCPU, RAM, NVMe, IPv4 and traffic included.

### Does Novita AI accept crypto?

No. Novita's FAQ states that cryptocurrency payments are not currently supported; it takes cards through Stripe and PayPal. GPUBank takes Bitcoin, USDT, Ethereum and more than a dozen other coins, and no cards.

### Why do Novita GPU prices differ between pages?

Novita's official pages disagree on some figures. The RTX 5090 is $0.56 on demand on the pricing page and console, and $0.73 on the Spot page; the RTX 4090 spot rate is $0.17 or $0.18. The docs say console prices are for reference only and the actual order prevails.

### Does Novita AI have serverless GPUs?

Yes. Novita sells serverless GPU endpoints with autoscaling and per-second worker billing, with prices shown only in the logged-in console. GPUBank has no serverless product and rents instances by the second.

### What happens to a stopped Novita instance?

Compute billing stops, but data is kept only 7 days at $0.005 per GB-day and then deleted, and pay-as-you-go hardware is released and may not be available on restart. On GPUBank a stopped instance keeps its persistent disk at $0.03 per GB-month.

### Are these prices with or without VAT?

Without. Every price on this page excludes VAT and sales tax, for Novita AI and for GPUBank, so the two columns compare like for like. Tax is added at checkout according to where the buyer is based.

## Sources

- https://novita.ai/pricing: GPU instance on-demand and spot prices, 1x and 8x, storage prices
- https://novita.ai/gpus: GPU price cross-check, platform claims (regions, Terraform, CLI, support, serverless)
- https://novita.ai/gpus-console/explore: Per-card vCPU/RAM, stock, Secure Cloud badges, on-demand prices
- https://novita.ai/gpus-spot: Spot prices, spot policy (1-hour protection, 1-hour notice)
- https://novita.ai/gpu-baremetal: Bare metal H100/B200 prices, H200/RTX configs, interconnect
- https://docs.novita.ai/guides/gpu-instance-pricing: Billing formula, per-second billing, storage prices
- https://docs.novita.ai/guides/gpu-instance-faq: Stopped instance retention, multi-GPU, subscriptions
- https://docs.novita.ai/guides/gpu-instance-overview: Free network transfer, templates, per-second billing
- https://docs.novita.ai/guides/gpu-instance-quickstart-manage-network-volume: Network volume billing and data center binding
- https://docs.novita.ai/guides/serverless-gpus-pricing: Serverless GPU billing method
- https://docs.novita.ai/guides/serverless-gpus-overview: Serverless features
- https://docs.novita.ai/api-reference/gpu-instance-create-instance.md: Billing modes (onDemand, monthly, spot), max 8 GPUs, VPC, network storage
- https://docs.novita.ai/api-reference/gpu-instance-list-products.md: Product fields incl. monthlyPrice, spotPrice, regions
- https://docs.novita.ai/guides/payment-methods: Stripe, minimum top-up, auto top-up
- https://docs.novita.ai/guides/faq: Payment methods (card, PayPal, no crypto), refunds, dedicated IP, restart behaviour
- https://docs.novita.ai/guides/quickstart: Sign-up methods, verification mention, Discord support
- https://novita.ai/user/register: Sign-up options, Sandbox credits
- https://docs.novita.ai/guides/gpu-instance-quickstart-connect-to-instance: SSH access

Prices change: check both providers before buying. Novita AI is a trademark of its owner; GPUBank is not affiliated with Novita AI.