GPUBank vs Spheron
Spheron is a GPU aggregator run by Beta360 Pte Ltd in Singapore. It resells on-demand, spot and reserved capacity from seven upstream providers, including Verda, Sesterce and Massed Compute, and their prices can change while an instance runs. On 2026-09-29 the lowest H100 SXM on-demand offer was $4.29 per GPU-hour, on 8-GPU nodes only; GPUBank charges $1.85 for a single H100 SXM. Spheron covers 12 countries and sells reserved clusters; GPUBank has fixed prices, per-second billing without a minimum and a stop state on every instance.
GPUBank
Tallinn, Estonia
H100 SXM
$1.85 / 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
Spheron
Singapore, Singapore
H100 SXM
$4.29 / h
- Capacity in 12 countries from seven upstream providers
- Reserved clusters of 8 to 512+ GPUs, with InfiniBand on some
- Spot H100 SXM from $2.23 per GPU-hour on 8-GPU nodes
Spheron
Aggregator
Singapore28 points compared26 of their pages readprices last checked on 29 September 2026page updated 2 November 2026
The verdict
Wider geography and cluster sourcing at Spheron. Fixed, lower prices and a stop state at GPUBank.
Spheron suits buyers who want one prepaid account over several second-tier GPU clouds, reserved clusters of 8 to 512+ GPUs, or capacity in 12 countries. Its prices are live offers set by upstream providers, not a list, and for the GPUs both of us rent GPUBank cost 38 to 68 % less than Spheron's lowest on-demand offer on the check date. Spheron charges a 20-minute minimum, and most of its providers bill the full rate until an instance is destroyed. Both include IPv4 and bandwidth and take stablecoins; Spheron also takes cards.
SpheronGPUBank 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 | Spheron per hour | GPUBank per hour | Difference | 100 h at Spheron | 100 h at GPUBank |
|---|---|---|---|---|---|
 B200 SXM180 GB |
$8.38 | $4.15 | −51 % | $838.00 | $415.00 |
 H200 SXM141 GB |
$4.80 | $2.49 | −48 % | $480.00 | $249.00 |
 H100 SXM80 GB |
$4.29 | $1.85 | −57 % | $429.00 | $185.00 |
 H100 PCIe80 GB |
$2.98 | $1.35 | −55 % | $298.00 | $135.00 |
 RTX PRO 600096 GB |
$2.35 | $1.15 | −51 % | $235.00 | $115.00 |
 A100 SXM80 GB |
$1.69 | $0.95 | −44 % | $169.00 | $95.00 |
 L40S48 GB |
$0.96 | $0.55 | −43 % | $96.00 | $55.00 |
 MI300X192 GB |
$3.59 | $1.65 | −54 % | $359.00 | $165.00 |
 RTX 509032 GB |
$0.78 | $0.48 | −38 % | $78.00 | $48.00 |
 RTX 409024 GB |
$0.72 | $0.23 | −68 % | $72.00 | $23.00 |
A fine-tune, 40 h on an H100 SXM
Spheron
$171.60
GPUBank
$74.00
Saves $97.60 at GPUBank (−57 %)
An inference box, a month, 720 h, on an L40S
Spheron
$691.20
GPUBank
$396.00
Saves $295.20 at GPUBank (−43 %)
Image generation, 200 h on an RTX 4090
Spheron
$144.00
GPUBank
$46.00
Saves $98.00 at GPUBank (−68 %)
02Their catalogue
Every GPU Spheron lists.
Spheron lists 30 GPU models. The 24 below are the ones recorded in our check, each at its lowest public price per GPU-hour. 10 of them are also rented at GPUBank (in blue).
- Blackwell and Hopper6
- B300262 GB · on demand$10.37
- B200192 GB · on demand$8.38
- H200141 GB · on demand$4.80
GH20096 GB · on demand$2.75- H100 SXM80 GB · on demand$4.29
- H100 PCIe80 GB · on demand$2.98
- Ampere, Ada and older data-centre10
- A100 SXM 80GB80 GB · on demand$1.69
A100 PCIe 80GB80 GB · on demand$1.43- A100 40GB40 GB · on demand$1.68
- Ampere, Ada and older data-centre continued10
- L40S48 GB · on demand$0.96
- L4048 GB · on demand$0.94
L424 GB · on demand$1.14
A4048 GB · on demand$2.23
A1024 GB · on demand$1.55
A1616 GB · on demand$0.61
V10032 GB · on demand$2.81- Workstation5
- RTX PRO 600096 GB · on demand$2.35
RTX 6000 Ada48 GB · on demand$0.78
- Workstation continued5
RTX A600048 GB · on demand$0.54
RTX A500024 GB · on demand$1.70
RTX A400016 GB · on demand$0.96- Consumer2
- RTX 509032 GB · on demand$0.78
- RTX 409024 GB · on demand$0.72
- AMD and other accelerators1
- MI300X192 GB · on demand$3.59
03What the price includes
What appears on the bill besides the GPU.
The same H100 SXM session of 2 h 14 m 03 s, line by line, as each provider describes its billing.
Spheron
2 h 14 m 03 s on H100 SXM
- GPU · $4.29 / h
- $9.58
- vCPU and RAM
- Included
- Storage
- Not published
- Egress
- Included
- Public IPv4
- Dedicated IPv4 included
- Billing granularity
- Per second in docs; per minute on the site
- Minimum charge
- 20 minutes; 60 on 2x MI300X
GPUBank
2 h 14 m 03 s on H100 SXM
- GPU · $1.85 / h
- $4.13
- 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.
Spheron
Most providers bill full rate until destroyed.
Stop exists only on Spheron AI (some types) and Spheron ES.
GPUBank
Only the persistent disk, $0.03 per GB-month. Compute, IPv4 and scratch cost nothing.
04Where the GPUs run
12 countries against 4, and where they are.
Countries where Spheron 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
Spheron12
United States
Canada
Finland
France
United Kingdom
Norway
Poland
Israel
India
Japan- Singapore
Australia
GPUBank4
- Paris
Frankfurt- Ashburn
- Singapore
Spheron · 12 countriesGPUBank · 4 regions
05Paying and signing up
How each one takes your money.
Spheron
Payments and sign-up
Payment methods
Cards
Invoice, enterprise onlyCrypto: USDC and USDT via CopperX
Minimum top-up $15
To start
- 1Email and password
- 2Email verification
Billing modelPrepaid credits
GPUBank
Payments and sign-up
Payment methods
No cardsA 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.
| Spheron | GPUBank | |
|---|---|---|
| SOC 2 | Not published | Not published |
| ISO 27001 | Not published | Not published |
| HIPAA | Not published | No |
| GDPR DPA | Not published | Published DPA, applies automatically |
| Availability target | No own SLA; provider SLAs up to 99.5 % | 99.9 % monthly, per region |
| Status page | None found | Public status page |
| Support | Chat, email, Discord; Slack for 100+ GPUs | 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 Spheron pricing page captured on 29 September 2026, and GPUBank's live board. Cookie banners and chat bubbles are closed; nothing else is edited.
- 1Per-GPU hourly rates are visible without an account, labelled cheapest live on-demand rate, and refresh every five minutes.
- 2The home page H100 headline of $2.98 matches an H100 PCIe offer; H100 SXM offers were priced higher.
- 3Pages carry named comparisons with Runpod, Lambda, CoreWeave, AWS, Azure and GCP, with Spheron shown cheaper.
- 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
28 points, read off both sites.
GPUBank aheadSpheron aheadEven
| Criterion | GPUBank | Spheron |
|---|---|---|
| Price and catalogue | ||
| H100 SXM, per GPU-hourLive offer from Verda on 2026-09-29, not a list price. The site headline H100 from $2.98 is an H100 PCIe offer. | $1.85, all-in | $4.29 lowest offer, 8-GPU nodes only |
| Cheapest GPU-hourSpot is lower: a V100 spot offer was $0.19 per GPU-hour. | $0.23, RTX 4090 | $0.54, RTX A6000 (lowest live offer) |
| GPU models listed20 models had live offers in the API; the rest are catalogue, reserve-only or provider-doc listings. | 10 | 30Ahead |
| What the GPU price includes | vCPU, RAM, NVMe, IPv4, traffic | vCPU, RAM, NVMe, IPv4, bandwidth |
| Billing | ||
| Billing granularityCharges settle on a five-minute cycle. | Per second | Per second in docs; per minute on the site |
| Minimum chargeSome machine types carry a longer provider minimum, charged in full on early termination. | None | 20 minutes; 60 on 2x MI300X |
| Stopped instanceStop exists only on Spheron AI (some types) and Spheron ES. | Disk only | Most providers bill full rate until destroyed |
| Persistent storageVolume rates are shown only in the authenticated API and dashboard. | $0.03 per GB-month | Not published |
| Egress traffic | Included | Included |
| Public IPv4 | Included | Dedicated IPv4 included |
| Billing model | Prepaid balance | Prepaid credits |
| Payment and sign-up | ||
| Crypto paymentsGPUBank takes more than a dozen coins, including Bitcoin and Ethereum. | Bitcoin, USDT, Ethereum and 12+ more | USDC and USDT via CopperX |
| Card payments | No | Cards via StripeAhead |
| Minimum deposit | $25 | $15Ahead |
| Platform | ||
| API and CLI | REST API, CLI, Python SDK | REST API only; keys need whitelisting |
| Terraform | Official provider | No |
| Serverless GPUs | No | No |
| Managed Kubernetes | No | No |
| Ready-made templates | 25 | 7 cloud-init startup scripts |
| Interruptible or spot | Yes, 40 % off | Yes, no SLA on spot |
| GPUs per instance | Up to 8 | Up to 10; clusters of 8 to 512+Ahead |
| Multi-node interconnect | Not offered | NVLink on SXM; InfiniBand on some clustersAhead |
| Network and regions | ||
| RegionsRegions belong to upstream providers; volumes cannot move between them. | 4 | 21 region entries across 7 providersAhead |
| Countries | France, Germany, US, Singapore | 12Ahead |
| EU regions | Paris, Frankfurt | Finland, France, PolandAhead |
| Trust and support | ||
| Availability targetMarketing pages cite 99.9 % to 99.99 %; the Terms say pass-through only. | 99.9 % monthly | No own SLA; provider SLAs up to 99.5 % |
| Compliance reports | None published | None of its own; partner certifications |
| Public status page | Yes | None found |
| Support | Tickets from a $25 balance | Chat, email, Discord; Slack for 100+ GPUs |
09The right fit
Different needs. A clear choice.
What Spheron does better
- Capacity in 12 countries from seven upstream providers
- Reserved clusters of 8 to 512+ GPUs, with InfiniBand on some
- Spot H100 SXM from $2.23 per GPU-hour on 8-GPU nodes
- 30 GPU models listed, including B300 and GH200
- Pay by card or USDC and USDT, from $15
Choose GPUBank if
- H100 SXM at $1.85 on a single GPU, against $4.29 on 8-GPU nodes
- Fixed prices, where Spheron rates can change on running instances
- No minimum runtime, where Spheron charges at least 20 minutes
- Stop any instance and pay only for its disk
- Terraform provider, CLI, Python SDK and 25 templates
Moving from Spheron takes an afternoon.
Spheron instances are VMs or bare metal with root SSH and a public IP, and no egress fee is published. Volumes are provider-specific and no snapshot or image export is documented.
- 1Launch the same GPU on GPUBank with a persistent disk in the region you need.
- 2Copy data from the Spheron instance with rsync or scp over its public IP.
- 3Run the same Docker image or pick the matching template, then destroy the Spheron instances.
# 1. a GPU and a disk on GPUBank gpubank launch --gpu h100-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 demand. Spheron's lowest H100 SXM on-demand offer on 2026-09-29 was $4.29 per GPU-hour, on 8-GPU nodes only. GPUBank charges $1.85 per GPU-hour for a single H100 SXM, with vCPU, RAM, NVMe, IPv4 and traffic included. Spheron spot H100 SXM started at $2.23, against $1.11 for interruptible on GPUBank.
No. Spheron resells capacity from seven upstream providers, and its terms allow rates to change on running instances without notice. Its site shows the cheapest live offer per GPU, refreshed every five minutes. GPUBank prices are fixed.
Only on some Spheron AI machine types and on Spheron ES. On Verda, Sesterce, Massed Compute, Spheron MS, Spheron AM, bare metal and spot, an instance bills the full hourly rate until destroyed. On GPUBank a stopped instance costs only its disk, at $0.03 per GB-month.
Yes. Spheron takes USDC and USDT through CopperX, and cards through Stripe, from $15 per deposit. GPUBank takes Bitcoin, USDT, Ethereum and more than a dozen other coins from $25, and no cards.
Not its own. Spheron's Terms pass through provider SLAs of up to 99.5 % on dedicated instances, and spot has none, while its marketing pages cite 99.9 % to 99.99 %. GPUBank states a 99.9 % monthly SLA target per region in its terms.
Without. Every price on this page excludes VAT and sales tax, for Spheron 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 Spheron was read on its own pages or official documentation. Prices last checked on 29 September 2026.
- 01spheron.network/positioning, headline rates, FAQ, stablecoins, compliance claims
- 02spheron.network/pricing/headline per-GPU rates, minimum runtime, payment FAQ, 99.9% SLA claim
- 03spheron.network/gpu-rental/per-GPU rates incl. last-known RTX 5090 / GH200 / MI300X, RTX 4090
- 04spheron.network/gpu-rental/h100/H100 2.98 headline, SXM5 and PCIe availability
- 05spheron.network/gpu-rental/mi300x/MI300X last-known 3.59, 1x / 2x configs, 60-minute minimum
- 06spheron.network/gpu-rental/gh200/GH200 3.02 last known
- 07spheron.network/gpu-rental/gb200/reserve-only NVL72
- 08spheron.network/features/on-demand-instances/per-minute billing, 99.99% SLA claim
- 09spheron.network/features/spot-instances/up to 50% off
- 10spheron.network/features/reserved-commitments/8 to 512+ GPUs, 1 to 12+ months
- 11spheron.network/terms/legal entity, pass-through SLA up to 99.5%, non-refundable credits, dynamic pricing, compliance under NDA
- 12spheron.network/privacy/entity
- 13app.spheron.ai/api/gpu-offers?limit=200&page=1all per-offer prices, configs, providers, regions (snapshot 11:39)
- 14docs.spheron.ai/billingcredits, per-second, minimum deposit, cards, USDC / USDT via CopperX, stopped billing
- 15docs.spheron.ai/concepts/regions-providersproviders, GPU tiers, regions, stop support, compliance per provider
- 16docs.spheron.ai/concepts/instance-typesspot / dedicated / bare metal, 99.95% SLA, minimum runtime
- 17docs.spheron.ai/concepts/instance-lifecyclestopped rate, stop support matrix
- 18docs.spheron.ai/concepts/networkingdedicated public IP, open ports
What we could not verify (7)
- Founding year, funding and headcount of Beta360 Pte Ltd / Spheron: not stated on the official pages read
- Volume storage prices per GB: only via the authenticated API or dashboard
- Reserved cluster floor prices and terms: shown in app.spheron.ai Reserved GPU, not captured; Terms say reserved is typically 12 months, docs say monthly terms
- Team discount levels (5 to 50% per API docs) are account specific
- No status page found (status.spheron.ai does not resolve)
- Prices are live marketplace values; the API snapshot of 2026-09-29 11:39 is one point in time and differed by about a dozen offers from a snapshot a few minutes earlier
- catalogue_count 30 = 20 models in the live API + RTX 4090, RTX 5090, GH200, MI300X (website) + GB200, GB300, R100 (reserve only) + Gaudi 2, H100 NVL, A30 (listed in provider docs, no live offer)
In short
GPUBank vs Spheron, 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
- Spheron: $4.29 lowest offer, 8-GPU nodes only
GPUBank: $1.85, all-in - Cheapest GPU-hour
- Spheron: $0.54, RTX A6000 (lowest live offer)
GPUBank: $0.23, RTX 4090 - Billing granularity
- Spheron: Per second in docs; per minute on the site
GPUBank: Per second - Stopped instance
- Spheron: Most providers bill full rate until destroyed
GPUBank: Disk only - Persistent storage
- Spheron: Not published
GPUBank: $0.03 per GB-month - Egress traffic
- Spheron: Included
GPUBank: Included - Crypto payments
- Spheron: USDC and USDT via CopperX
GPUBank: Bitcoin, USDT, Ethereum and 12+ more - Card payments
- Spheron: Cards via Stripe
GPUBank: No - Minimum deposit
- Spheron: $15
GPUBank: $25 - GPU models listed
- Spheron: 30
GPUBank: 10 - Regions
- Spheron: 21 region entries across 7 providers
GPUBank: 4 - Serverless GPUs
- Spheron: No
GPUBank: No - Availability target
- Spheron: No own SLA; provider SLAs up to 99.5 %
GPUBank: 99.9 % monthly - Compliance reports
- Spheron: None of its own; partner certifications
GPUBank: None published - Support
- Spheron: Chat, email, Discord; Slack for 100+ GPUs
GPUBank: Tickets from a $25 balance - Headquarters
- Spheron: Singapore, Singapore
GPUBank: Tallinn, Estonia
How this page was made
Every price and policy of Spheron 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. Spheron, its logo and its website belong to their owner and are shown only to identify the service compared. GPUBank is not affiliated with Spheron.
