GPUBank vs Beam
Beam, run by Smartshare, Inc. and backed by Y Combinator, is a US serverless platform for GPU endpoints, task queues and sandboxes, with on-demand GPU machines and reserved clusters on the side. It lists no H100 SXM outside contact-sales clusters; its H100 PCIe machine starts at $1.83 per hour, and a serverless H100 PCIe costs $0.000972 per second ($3.50 per hour) for the GPU alone. GPUBank charges $1.35 for an H100 PCIe and $1.85 for an H100 SXM, with vCPU, RAM, NVMe and traffic included.
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
Beam
Cambridge, United States
H200 SXM
$2.09 / h
- Serverless GPU endpoints and task queues that scale to zero
- H200 from $2.09 and RTX PRO 6000 from $1.09 on demand
- Storage free up to 1 TB, then $0.021 per GB-month
Beam
Serverless
United States25 points compared10 of their pages readprices last checked on 29 September 2026page updated 2 November 2026
The verdict
Serverless scale-to-zero at Beam. Lower prices, named regions and crypto top-ups at GPUBank.
Beam fits Python teams that deploy inference endpoints or task queues through its SDK and want them to scale to zero. Its on-demand machines are sold at 'from' prices: H200 from $2.09, RTX PRO 6000 from $1.09 and B200 from $4.11 are at or below GPUBank. On the H100 PCIe, A100, L40S, RTX 5090 and RTX 4090, GPUBank costs 26 to 48 % less. Beam takes cards through Stripe, holds SOC 2 Type II and asks you to contact it to choose a region; GPUBank takes crypto and lets you pick Paris, Frankfurt, Ashburn or Singapore.
BeamGPUBank 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 | Beam per hour | GPUBank per hour | Difference | 100 h at Beam | 100 h at GPUBank |
|---|---|---|---|---|---|
 B200 SXM180 GB |
$4.11 | $4.15 | +1 % | $411.00 | $415.00 |
 H200 SXM141 GB |
$2.09 | $2.49 | +19 % | $209.00 | $249.00 |
 H100 PCIe80 GB |
$1.83 | $1.35 | −26 % | $183.00 | $135.00 |
 RTX PRO 600096 GB |
$1.09 | $1.15 | +6 % | $109.00 | $115.00 |
 A100 SXM80 GB |
$1.36 | $0.95 | −30 % | $136.00 | $95.00 |
 L40S48 GB |
$0.76 | $0.55 | −28 % | $76.00 | $55.00 |
 RTX 509032 GB |
$0.72 | $0.48 | −33 % | $72.00 | $48.00 |
 RTX 409024 GB |
$0.44 | $0.23 | −48 % | $44.00 | $23.00 |
Beam's figures are 'from' prices for on-demand machines with vCPU, RAM and NVMe included; the GPU count per machine is not stated. Serverless GPU rates are higher and exclude CPU and RAM.
An inference box, a month, 720 h, on an L40S
Beam
$547.20
GPUBank
$396.00
Saves $151.20 at GPUBank (−28 %)
Image generation, 200 h on an RTX 4090
Beam
$88.00
GPUBank
$46.00
Saves $42.00 at GPUBank (−48 %)
A long-context job, 40 h on an H200
Beam
$83.60
GPUBank
$99.60
Costs $16.00 more at GPUBank (+19 %)
02Their catalogue
Every GPU Beam lists.
Beam lists 13 GPU models. The 9 below are the ones recorded in our check, each at its lowest public price per GPU-hour. 8 of them are also rented at GPUBank (in blue).
- B200180 GB · on demand$4.11
- H200141 GB · on demand$2.09
- H100 PCIe80 GB · on demand$1.83
- A100 SXM 80GB80 GB · on demand$1.36
- L40S48 GB · on demand$0.76
- RTX PRO 600096 GB · on demand$1.09
RTX A600048 GB · on demand$0.54- RTX 509032 GB · on demand$0.72
- RTX 409024 GB · on demand$0.44
At GPUBank
vCPU, RAM, NVMe and traffic included. All prices
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.
Beam
2 h 14 m 03 s on H200 SXM
- GPU · $2.09 / h
- $4.67
- vCPU and RAM
- Billed separately
- Storage
- Free to 1 TB, then $0.021 per GB-month
- Egress
- Included
- Public IPv4
- Not published
- Billing granularity
- By the millisecond (serverless)
- 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.
Beam
Scales to zero; keep-warm time is billed.
Default keep-warm: 180 s for endpoints, 10 s for task queues, 600 s for pods.
GPUBank
Only the persistent disk, $0.03 per GB-month. Compute, IPv4 and scratch cost nothing.
04Where the GPUs run
1 country against 4, and where they are.
Countries where Beam 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.
United States
Paris
Frankfurt
Ashburn
Singapore
Beam1
- United States
GPUBank4
Paris
Frankfurt- Ashburn
Singapore
Beam · 1 countryGPUBank · 4 regions
05Paying and signing up
How each one takes your money.
Beam
Payments and sign-up
Payment methods
Cards
No cryptoMinimum top-up Not published
To start
- 1Email
- 2Card, valid payment method
Billing modelBilled in arrears; prepaid credits where offered
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.
| Beam | GPUBank | |
|---|---|---|
| SOC 2 | Type II | Not published |
| ISO 27001 | Not published | Not published |
| HIPAA | BAA available | No |
| GDPR DPA | DPA available | Published DPA, applies automatically |
| Availability target | No SLA published | 99.9 % monthly, per region |
| Status page | status.beam.cloud | Public status page |
| Support | Community; live chat from the $89/month plan | 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 Beam pricing page captured on 29 September 2026, and GPUBank's live board. Cookie banners and chat bubbles are closed; nothing else is edited.
- 1The pricing page shows serverless per-second GPU rates, on-demand 'from' hourly prices and cluster rows marked Contact us.
- 2The headline 'H100s from $1.83/hr' refers to the on-demand H100 PCIe machine, not the SXM.
- 3A cost calculator works without an account, and serverless rates toggle between per second and per hour.
- 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
25 points, read off both sites.
GPUBank aheadBeam aheadEven
| Criterion | GPUBank | Beam |
|---|---|---|
| Price and catalogue | ||
| H100 SXM, per GPU-hourCompared with our H100 PCIe at $1.35. Beam's H100 SXM is sold only in reserved clusters, price on request. | $1.85, all-in | No H100 SXM on demand; H100 PCIe from $1.83 |
| Cheapest GPU-hour | $0.23, RTX 4090 | $0.44, RTX 4090 machine (from) |
| GPU models listed | 10 | 13Ahead |
| What the GPU price includesServerless CPU and RAM are billed on top per second. | vCPU, RAM, NVMe, IPv4, traffic | Machines: vCPU, RAM, NVMe; serverless: GPU only |
| Billing | ||
| Billing granularity | Per second | By the millisecond (serverless)Ahead |
| Minimum charge | None | Not published |
| Stopped instanceDefault keep-warm: 180 s for endpoints, 10 s for task queues, 600 s for pods. | Disk only | Scales to zero; keep-warm time is billed |
| Persistent storage | $0.03 per GB-month | Free to 1 TB, then $0.021 per GB-monthAhead |
| Egress traffic | Included | Included |
| Public IPv4 | Included | Not published |
| Billing model | Prepaid balance | Billed in arrears; prepaid credits where offered |
| Payment and sign-up | ||
| Crypto paymentsBeam's terms name Stripe card payments; PayPal, wire and crypto are not mentioned. | Bitcoin, USDT, Ethereum and 12+ more | Not listed: cards via Stripe only |
| Card payments | No | Cards via StripeAhead |
| Minimum deposit | $25 | Not published |
| Platform | ||
| API and CLI | REST API, CLI, Python SDK | Python and TypeScript SDKs, CLI, REST API |
| Terraform | Official provider | None documented |
| Serverless GPUs | No | Yes, the core productAhead |
| Managed Kubernetes | No | No |
| Ready-made templates | 25 | Not published |
| Interruptible or spot | Yes, 40 % off | No |
| GPUs per instance | Up to 8 | By request; 8-GPU nodes in reserved clustersAhead |
| Multi-node interconnect | Not offered | InfiniBand on reserved clustersAhead |
| Network and regions | ||
| RegionsWorkloads run in the US unless you contact Beam to choose another region. | 4 | US, Europe, Asia; no locations named |
| Countries | France, Germany, US, Singapore | United States is the only one named |
| EU regions | Paris, Frankfurt | Europe stated, on request, location not named |
| Trust and support | ||
| Availability target | 99.9 % monthly | No SLA published |
| Compliance reports | None published | SOC 2 Type II; HIPAA BAA on EnterpriseAhead |
| Public status page | Yes | status.beam.cloud |
| Support | Tickets from a $25 balance | Community; live chat from the $89/month plan |
09The right fit
Different needs. A clear choice.
What Beam does better
- Serverless GPU endpoints and task queues that scale to zero
- H200 from $2.09 and RTX PRO 6000 from $1.09 on demand
- Storage free up to 1 TB, then $0.021 per GB-month
- SOC 2 Type II, HIPAA BAA on Enterprise
- Reserved 8-GPU SXM clusters with InfiniBand, open-source beta9 engine
Choose GPUBank if
- H100 PCIe at $1.35 against $1.83, RTX 4090 at $0.23 against $0.44
- H100 SXM on demand at $1.85, which Beam sells only in clusters
- Firm published prices, not 'from' figures
- Pay in Bitcoin, USDT and 12+ other coins, no card
- Four named regions you pick yourself, plus a Terraform provider
Moving from Beam takes an afternoon.
Beam charges no egress, so moving data out is free on its side. Code written with Beam SDK decorators has to be rewritten as plain containers, while custom Docker images and volumes carry over.
- 1Launch the same GPU on GPUBank with a persistent disk in the region you need.
- 2Copy volume data with rsync or rclone, then run your existing Docker image.
- 3Replace endpoint and task_queue decorators with a web server or worker in the container.
# 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 for the H100 PCIe: Beam's on-demand machine starts at $1.83 per hour and GPUBank charges $1.35. Beam's serverless H100 PCIe costs $0.000972 per second, about $3.50 per hour, before CPU and RAM. Beam sells the H100 SXM only in reserved clusters, while GPUBank rents it at $1.85 per GPU-hour.
Yes. Serverless GPU rates are for the GPU only; CPU costs $0.000105 per second per core and RAM $0.0000055 per second per GiB. Beam's own example of an RTX 4090 endpoint with 2 cores and 16 GiB comes to $1.77 per hour. GPUBank includes vCPU, RAM and NVMe in the GPU price.
Beam's terms name card payments through Stripe and do not mention crypto. GPUBank is paid from a USD balance topped up in Bitcoin, USDT, Ethereum and more than a dozen other coins, from $25.
Beam. Its endpoints and task queues start containers on demand and bill by the millisecond while they run, with a default keep-warm of 180 s on endpoints. GPUBank has no serverless product; you rent a VM and pay per second until you stop it.
Beam states primary locations in the United States, Europe and Asia, runs on AWS and Google Cloud, and uses the United States unless you contact it to pick another region. GPUBank runs in Paris, Frankfurt, Ashburn and Singapore, chosen at launch.
Without. Every price on this page excludes VAT and sales tax, for Beam 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 Beam was read on its own pages or official documentation. Prices last checked on 29 September 2026.
- 01beam.cloud/pricingServerless, sandbox, on-demand, cluster, storage, BYOC prices, plans, egress
- 02docs.beam.cloud/v2/environment/gpuServerless GPU list, on-demand GPU list, multi-GPU by request, regions
- 03docs.beam.cloud/v2/resources/pricing-and-billingWhat is billed, keep-warm defaults, storage
- 04docs.beam.cloud/v2/security/securitySOC 2 Type II, HIPAA BAA, DPA, AWS/GCP infrastructure, default US region
- 05docs.beam.cloud/v2/security/terms-and-conditionsLegal entity, Stripe card payments, prepaid credits, billing in arrears, suspension
- 06docs.beam.cloud/v2/self-hosting/overviewOpen-source beta9
- 07docs.beam.cloud/v2/reference/cliCLI, machine list
- 08status.beam.cloud/Status page and 90-day uptime
- 09beam.cloud/Y Combinator backing, SDK languages, positioning
- 10beam.cloud/aboutCompany description
What we could not verify (8)
- Serverless prices for T4 and A10G (listed as available serverless in docs) not published on the pricing page.
- Number of GPUs per on-demand machine and the non-'from' price range not published.
- Cluster (H100/H200/B200/B300 SXM) prices: contact sales.
- Payment methods beyond card via Stripe (PayPal, wire, crypto) not published.
- Regions: only 'United States, Europe, and Asia' stated; specific countries and cities not published.
- SLA not published. Terraform and IPv4 policy not published.
- Minimum deposit and free credits for new accounts not published on pages read.
- HQ city and founding year not published on pages read.
In short
GPUBank vs Beam, 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
- Beam: No H100 SXM on demand; H100 PCIe from $1.83
GPUBank: $1.85, all-in - Cheapest GPU-hour
- Beam: $0.44, RTX 4090 machine (from)
GPUBank: $0.23, RTX 4090 - Billing granularity
- Beam: By the millisecond (serverless)
GPUBank: Per second - Stopped instance
- Beam: Scales to zero; keep-warm time is billed
GPUBank: Disk only - Persistent storage
- Beam: Free to 1 TB, then $0.021 per GB-month
GPUBank: $0.03 per GB-month - Egress traffic
- Beam: Included
GPUBank: Included - Crypto payments
- Beam: Not listed: cards via Stripe only
GPUBank: Bitcoin, USDT, Ethereum and 12+ more - Card payments
- Beam: Cards via Stripe
GPUBank: No - Minimum deposit
- Beam: Not published
GPUBank: $25 - GPU models listed
- Beam: 13
GPUBank: 10 - Regions
- Beam: US, Europe, Asia; no locations named
GPUBank: 4 - Serverless GPUs
- Beam: Yes, the core product
GPUBank: No - Availability target
- Beam: No SLA published
GPUBank: 99.9 % monthly - Compliance reports
- Beam: SOC 2 Type II; HIPAA BAA on Enterprise
GPUBank: None published - Support
- Beam: Community; live chat from the $89/month plan
GPUBank: Tickets from a $25 balance - Headquarters
- Beam: Cambridge, United States
GPUBank: Tallinn, Estonia
How this page was made
Every price and policy of Beam 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. Beam, its logo and its website belong to their owner and are shown only to identify the service compared. GPUBank is not affiliated with Beam.
