GPUBank vs Sesterce
Sesterce is a French AI-infrastructure company from Marseille that builds AI factories in France for contract buyers and runs Sesterce Cloud, a self-serve console of on-demand GPU VMs in 22 console regions. The console had no H100 SXM on the check date: its H100 PCIe offers ranged from $3.00 to $6.59 per hour, against $1.35 for an H100 PCIe and $1.85 for an H100 SXM at GPUBank. Sesterce lists more GPU models, countries and certifications; GPUBank has fixed prices, per-second billing and crypto top-ups.
GPUBank
Tallinn, Estonia
H200 SXM
$2.49 / h
- vCPU, RAM, NVMe, IPv4 and traffic included
- Per-second billing after the health check
- Top up in Bitcoin, USDT and 12+ other coins
vCPU, RAM, NVMe, IPv4 and traffic included
Sesterce
Marseille, France
H200 SXM
$2.48 / h
- 21 GPU models, 17 of them bookable in the console
- RTX A4000 from $0.17 per hour, a card GPUBank does not list
- Autoscaling inference endpoints on L40S, H100 or A100
Sesterce
European cloud
France23 points compared22 of their pages readprices last checked on 29 September 2026page updated 2 November 2026
The verdict
More models, inference endpoints and certifications at Sesterce. Lower, fixed prices and crypto top-ups at GPUBank.
Sesterce suits buyers who want to compare many mid-range GPU offers in one console, pay by card or bank transfer, or sign a contract for French AI-factory capacity with SOC 2 and ISO 27001 claims. Its console prices are live offers that moved within minutes during the check, and no H100 SXM, H200 or B200 offer was bookable. On the five other models in the price table, GPUBank costs 28 to 65 % less per GPU-hour; the homepage from prices for H200 and B200 sit one and four cents below GPUBank but were not bookable. GPUBank bills by the second, publishes its disk, IPv4 and traffic terms, and takes crypto; Sesterce bills by the minute and publishes none of those extras.
SesterceGPUBank 01The real cost
The same GPUs, side by side, per hour.
Every GPU model both of us rent, at each provider's own published price for one card, in US dollars, excluding VAT. Prices last checked on 29 September 2026.
| GPU | Sesterce per hour | GPUBank per hour | Difference | 100 h at Sesterce | 100 h at GPUBank |
|---|---|---|---|---|---|
 B200 SXM180 GB |
$4.11 | $4.15 | +1 % | $411.00 | $415.00 |
 H200 SXM141 GB |
$2.48 | $2.49 | −0 % | $248.00 | $249.00 |
 H100 PCIe80 GB |
$3.00 | $1.35 | −55 % | $300.00 | $135.00 |
 RTX PRO 600096 GB |
$2.41 | $1.15 | −52 % | $241.00 | $115.00 |
 A100 SXM80 GB |
$1.52 | $0.95 | −38 % | $152.00 | $95.00 |
 L40S48 GB |
$0.76 | $0.55 | −28 % | $76.00 | $55.00 |
 RTX 409024 GB |
$0.66 | $0.23 | −65 % | $66.00 | $23.00 |
Sesterce prices are live console offers per instance-hour, shown per GPU, that changed within minutes. B200, H200 and L40S figures are homepage from prices; no B200 or H200 offer was bookable.
An inference box, a month, 720 h, on an L40S
Sesterce
$547.20
GPUBank
$396.00
Saves $151.20 at GPUBank (−28 %)
Image generation, 200 h on an RTX 4090
Sesterce
$132.00
GPUBank
$46.00
Saves $86.00 at GPUBank (−65 %)
A long-context job, 40 h on an H200
Sesterce
$99.20
GPUBank
$99.60
Costs $0.40 more at GPUBank (+0 %)
02Their catalogue
Every GPU Sesterce lists.
21 GPU models listed by Sesterce, each at its lowest public price per GPU-hour. 7 of them are also rented at GPUBank (in blue).
- Blackwell and Hopper5
- B300on demand$4.95
- B200on demand$4.11
- H200on demand$2.48
- H100 PCIe80 GB · on demand$3.00
- H10080 GB · on demand · variant not stated$1.83
- Ampere, Ada and older data-centre10
- A100 SXM 80GB80 GB · on demand$1.52
A100 PCIe 80GB80 GB · on demand$1.49
- Ampere, Ada and older data-centre continued10
- A100 40GB40 GB · on demand$1.54
- L40S48 GB · on demand$0.76
- L4048 GB · on demand$0.97
L424 GB · on demand$1.05
A4048 GB · on demand$2.05
A1024 GB · on demand$1.42
A1616 GB · on demand$0.56
V10032 GB · on demand$2.57
- Workstation5
- RTX PRO 600096 GB · on demand$2.41
RTX 6000 Ada48 GB · on demand$1.07
RTX A600048 GB · on demand$0.63
RTX A500024 GB · on demand$1.56
RTX A400016 GB · on demand$0.17- Consumer1
- RTX 409024 GB · on demand$0.66
03What the price includes
What appears on the bill besides the GPU.
The same H200 SXM session of 2 h 14 m 03 s, line by line, as each provider describes its billing.
Sesterce
2 h 14 m 03 s on H200 SXM
- GPU · $2.48 / h
- $5.54
- vCPU and RAM
- Included
- Storage
- Not published
- Egress
- Not published
- Public IPv4
- Not published
- Billing granularity
- Per minute
- Minimum charge
- Not published
GPUBank
2 h 14 m 03 s on H200 SXM
- GPU · $2.49 / h
- $5.56
- vCPU and RAM
- Included
- Storage
- Local NVMe included; persistent disk $0.03 per GB-month
- Egress
- Included
- Public IPv4
- Included
- Billing granularity
- Per second, from the health check
- Minimum charge
- None
While stopped
What keeps running on the bill.
Sesterce
Not published.
Docs describe a Stop action but not how a stopped instance is billed.
GPUBank
Only the persistent disk, $0.03 per GB-month. Compute, IPv4 and scratch cost nothing.
04Where the GPUs run
7 countries against 4, and where they are.
Countries where Sesterce says its GPUs run, from its own region list or documentation, against GPUBank's four regions. A pin marks a country, not the exact site.
Paris
Frankfurt
Ashburn
Singapore
Sesterce7
Canada- France
India
Israel
Norway
Poland
United States
GPUBank4
- Paris
Frankfurt- Ashburn
Singapore
Sesterce · 7 countriesGPUBank · 4 regions
05Paying and signing up
How each one takes your money.
Sesterce
Payments and sign-up
Payment methods
Cards
Bank transfer
No cryptoMinimum top-up Not published
To start
- 1Email or GitHub/Google OAuth
Billing modelPrepaid balance, debited every minute
GPUBank
Payments and sign-up
Payment methods
A prepaid USD balance, topped up in crypto. Every GPU-second is drawn from it.
Minimum top-up $25
To start
- 1Email and password
- 2Top up in crypto
- 3Launch a GPU
Billing modelPrepaid balance
06Trust and support
Certificates, uptime and who answers.
| Sesterce | GPUBank | |
|---|---|---|
| SOC 2 | Not published | Not published |
| ISO 27001 | Listed, report not public | Not published |
| HIPAA | Not published | No |
| GDPR DPA | DPA available | Published DPA, applies automatically |
| Availability target | 99.99% cluster SLA claimed, text not linked | 99.9 % monthly, per region |
| Status page | None; status.sesterce.com did not resolve | Public status page |
| Support | Request forms; 24/7 NOC claimed | Tickets from a $25 balance |
07Their site, our site
What you see before you pay.
Both sites as a visitor sees them, without an account: the Sesterce home page captured on 29 September 2026, and GPUBank's live board. Cookie banners and chat bubbles are closed; nothing else is edited.
- 1The /pricing URL returns 404; Pricing in the menu is an anchor on the homepage, which shows two price blocks that do not agree.
- 2The console offer list and checkout panels are public, priced per instance-hour, so multi-GPU prices must be divided by the GPU count.
- 3Ticker prices changed between page loads, and the docs carry an older price table that does not match the console.
- 1Live stock per region and every price, without an account.
- 2Per-hour and per-second prices, with vCPU, RAM, NVMe, IPv4 and traffic in them.
- 3A machine-readable price feed at /prices.json.
08The spec sheet
23 points, read off both sites.
GPUBank aheadSesterce aheadEven
| Criterion | GPUBank | Sesterce |
|---|---|---|
| Price and catalogue | ||
| H100 SXM, per GPU-hourCompared with GPUBank's H100 PCIe at $1.35. The homepage shows H100 from $1.83, mode and form factor not stated. | $1.85, all-in | No H100 SXM; H100 PCIe $3.00 to $6.59 |
| Cheapest GPU-hour$0.165 per GPU-hour on 2x and 4x RTX A4000 VMs in NOR-Oslo-1. | $0.23, RTX 4090 | $0.17, RTX A4000 (1x)Ahead |
| GPU models listed | 10 | 21 (17 bookable in the console)Ahead |
| What the GPU price includesSesterce does not publish its egress or IPv4 terms; GPUBank includes both. | vCPU, RAM, NVMe, IPv4, traffic | vCPU, RAM, non-persistent disk |
| Billing | ||
| Billing granularity | Per second | Per minute |
| Minimum charge | None | Not published |
| Stopped instanceDocs describe a Stop action but not how a stopped instance is billed. | Disk only | Not published |
| Persistent storagePersistent volumes exist only in some regions and must be attached at launch. | $0.03 per GB-month | Not published |
| Egress traffic | Included | Not published |
| Public IPv4 | Included | Not published |
| Billing modelAt zero balance, docs say instances are stopped and storage volumes deleted. | Prepaid balance | Prepaid balance, debited every minute |
| Payment and sign-up | ||
| Crypto payments | Bitcoin, USDT, Ethereum and 12+ more | Not offered |
| Card payments | No | Cards via Stripe, bank transfer, walletsAhead |
| Minimum deposit | $25 | Not published |
| Platform | ||
| API and CLI | REST API, CLI, Python SDK | REST API; CLI claimed, not documented |
| Terraform | Official provider | Claimed on homepage, no docs found |
| Serverless GPUs | No | Autoscaling inference endpointsAhead |
| Managed KubernetesNo managed Kubernetes in the self-serve console. | No | On contract clusters (Sesterce OS)Ahead |
| Ready-made templates | 25 | 6 launch images |
| Interruptible or spotA Spot filter exists but returned no offers; docs say only on-demand pricing exists. | Yes, 40 % off | No |
| GPUs per instance | Up to 8 | Up to 10 per VM, contract clustersAhead |
| Multi-node interconnect | Not offered | NVLink on some 8x A100; InfiniBand on contractAhead |
| Network and regions | ||
| RegionsThe operator behind each console region is not disclosed. | 4 | 22 in the consoleAhead |
| Countries | France, Germany, US, Singapore | US, Canada, Norway, Poland, France, India, IsraelAhead |
| EU regions | Paris, Frankfurt | Paris and Warsaw, plus Oslo |
| Trust and support | ||
| Availability targetThe Trust page's SLA link points nowhere. GPUBank's 99.9 % target is in its terms. | 99.9 % monthly | 99.99% cluster SLA claimed, text not linked |
| Compliance reportsSOC 2 type not stated, report under NDA; no certificates linked. | None published | SOC 2, ISO 27001, ISO 42001, HDS claimedAhead |
| Public status page | Yes | None; status.sesterce.com did not resolve |
| Support | Tickets from a $25 balance | Request forms; 24/7 NOC claimedAhead |
09The right fit
Different needs. A clear choice.
What Sesterce does better
- 21 GPU models, 17 of them bookable in the console
- RTX A4000 from $0.17 per hour, a card GPUBank does not list
- Autoscaling inference endpoints on L40S, H100 or A100
- Cards, bank transfer, Google Pay, Cash App Pay and Link
- SOC 2, ISO 27001 and HDS claims, contract clusters in French AI factories
Choose GPUBank if
- H100 PCIe at $1.35, against $3.00 to $6.59 in the Sesterce console
- H100 SXM on demand at $1.85; Sesterce had none bookable
- Fixed published prices, not live offers that change within minutes
- Per-second billing, interruptible and reserved tiers
- Pay in Bitcoin, USDT and 12+ other coins
Moving from Sesterce takes an afternoon.
Sesterce instance disks are non-persistent and no volume export or snapshot is documented, so data leaves over SSH. Workloads run on standard images such as vLLM and PyTorch, which exist on GPUBank too.
- 1Launch the same GPU on GPUBank with a persistent disk in the region you need.
- 2Copy data from the Sesterce instance or volume with rsync over SSH before you stop it.
- 3Run the same public Docker image or pick the matching template, then keep your Sesterce balance above zero until the copy ends.
# 1. a GPU and a disk on GPUBank gpubank launch --gpu h200-sxm --region eu-paris --disk data # 2. copy your data from the old machine rsync -azP old-host:/workspace/ root@<new-ip>:/workspace/ # 3. run the same image or template ssh root@<new-ip> 'cd /workspace && ./start.sh'
A few things you might be wondering.
Yes, on the check date. The only bookable H100 offers in the Sesterce console were PCIe, from $3.00 to $6.59 per hour. GPUBank charges $1.35 for an H100 PCIe and $1.85 for an H100 SXM, with vCPU, RAM, NVMe, IPv4 and traffic included.
The homepage widget shows a from price per GPU-hour for H100 SXM5 or PCIe without naming the form factor. It also shows $1.66 in a CLI block with no stated mode. Neither price was bookable in the console during the check.
No. Its docs list cards through Stripe, bank transfer, Cash App Pay, Google Pay and Link. GPUBank is paid from a USD balance topped up in Bitcoin, USDT, Ethereum and more than a dozen other coins, and takes no cards.
Its homepage lists H200 from $2.48 and B200 from $4.11 per GPU-hour, but no H200 or B200 offer was bookable in the console on the check date. Those homepage figures are slightly below GPUBank, which rents the H200 SXM at $2.49 and the B200 SXM at $4.15 on demand.
Sesterce, for contract buyers. It runs AI factories in Valence and Paris and claims SOC 2, ISO 27001 and HDS, though no certificates are linked. GPUBank is a French company with regions in Paris and Frankfurt and publishes no SOC 2 or ISO 27001 report.
Without. Every price on this page excludes VAT and sales tax, for Sesterce and for GPUBank, so the two columns compare like for like. Tax is added at checkout according to where the buyer is based.
Sources
Every figure about Sesterce was read on its own pages or official documentation. Prices last checked on 29 September 2026.
- 01sesterce.com/sites, silicon, CLI price block, live-offers widget prices, capacity request tiers
- 02sesterce.com/pricingsite observation: registry pricing URL is dead
- 03cloud.sesterce.com/computepublic offer list, ticker, filters (Spot, Bare-Metal, Persistent Volume)
- 04cloud.sesterce.com/compute/new?gpuType=<GPU>&numGpus=<N>per-region offer prices, vCPU, RAM, storage, form factor, NVLink (raw/extra-compute-checkout-all.txt, raw/extra-compute-offers-by-config.txt)
- 05cloud.sesterce.com/clusterscluster booking by request
- 06docs.sesterce.com/older docs price table, product description
- 07docs.sesterce.com/welcome-on-sesterce-cloud/payment-and-billingon-demand only, prepaid credits, payment methods, per-minute debit, zero-balance policy
- 08docs.sesterce.com/welcome-on-sesterce-cloud/invoicingtop-up and monthly invoices
- 09docs.sesterce.com/welcome-on-sesterce-cloud/account-creationsign-up via email or GitHub/Google OAuth, GDPR statement
- 10docs.sesterce.com/compute-instancesnon-persistent pod storage, volume filter
- 11docs.sesterce.com/compute-instances/configure-your-compute-instanceregions, volumes at launch, OS images, SSH keys
- 12docs.sesterce.com/compute-instances/configure-your-compute-instance/persistent-storage-volumesvolume rules (region/AZ-bound)
- 13docs.sesterce.com/manage-your-instancesstatuses, no billing while Pending, Stop action
- 14docs.sesterce.com/api-referenceAPI scope, API key prerequisites
- 15docs.sesterce.com/api-reference/gpu-cloud-instancesoffers endpoint, deploymentType vm/container/baremetal
- 16docs.sesterce.com/api-reference/volumesvolume minimum size (10 GB), price field shown only as example
- 17docs.sesterce.com/ai-inference-instancesinference product, edge nodes, model catalogue
- 18docs.sesterce.com/ai-inference-instances/inference-instance-configuration/select-your-flavorinference GPUs L40S, H100, A100
What we could not verify (11)
- Founding year and funding amounts: not stated on the pages read.
- Persistent volume price, egress price, IPv4 policy, minimum top-up and free credits: not published.
- Billing of stopped instances: not documented.
- Whether VAT is added to console prices: not stated.
- Which operator runs each console region (e.g. USA-NewYork-1, CAN-Montreal-2): not disclosed.
- SOC 2 report type (Type I or II): not stated; certificates and SLA text were not accessible (links inactive).
- CLI and Terraform provider: claimed on the homepage but no documentation found on docs.sesterce.com.
- Prices for reserved capacity, clusters, AI-factory pods and inference endpoints: not published.
- Mode of the homepage CLI-section prices (B200 4.50, H200 2.25, H100 1.66, A100 1.24): not stated, so not recorded as price rows.
- H100 SXM on-demand price: no bookable H100 SXM offer in the console; the homepage 1.83 'from' price does not say SXM or PCIe.
- VRAM of B300, B200 and H200 as offered: not stated on the homepage widget.
In short
GPUBank vs Sesterce, in facts.
The essentials of this comparison in one table, for quick readers and AI assistants. Prices last checked on 29 September 2026.
Read as Markdown- H100 SXM, per GPU-hour
- Sesterce: No H100 SXM; H100 PCIe $3.00 to $6.59
GPUBank: $1.85, all-in - Cheapest GPU-hour
- Sesterce: $0.17, RTX A4000 (1x)
GPUBank: $0.23, RTX 4090 - Billing granularity
- Sesterce: Per minute
GPUBank: Per second - Stopped instance
- Sesterce: Not published
GPUBank: Disk only - Persistent storage
- Sesterce: Not published
GPUBank: $0.03 per GB-month - Egress traffic
- Sesterce: Not published
GPUBank: Included - Crypto payments
- Sesterce: Not offered
GPUBank: Bitcoin, USDT, Ethereum and 12+ more - Card payments
- Sesterce: Cards via Stripe, bank transfer, wallets
GPUBank: No - Minimum deposit
- Sesterce: Not published
GPUBank: $25 - GPU models listed
- Sesterce: 21 (17 bookable in the console)
GPUBank: 10 - Regions
- Sesterce: 22 in the console
GPUBank: 4 - Serverless GPUs
- Sesterce: Autoscaling inference endpoints
GPUBank: No - Availability target
- Sesterce: 99.99% cluster SLA claimed, text not linked
GPUBank: 99.9 % monthly - Compliance reports
- Sesterce: SOC 2, ISO 27001, ISO 42001, HDS claimed
GPUBank: None published - Support
- Sesterce: Request forms; 24/7 NOC claimed
GPUBank: Tickets from a $25 balance - Headquarters
- Sesterce: Marseille, France
GPUBank: Tallinn, Estonia
How this page was made
Every price and policy of Sesterce was read on its own website, documentation or public API on 29 September 2026, with the address of each page listed above. A value we could not read on a primary source is marked "Not published" rather than estimated. GPUBank figures come from this site and its terms. Prices change: check both providers before you buy. Sesterce, its logo and its website belong to their owner and are shown only to identify the service compared. GPUBank is not affiliated with Sesterce.
