Beam Cloud review
Updated September 28, 2026 · DeployCue Editorial Team · Facts checked September 26, 2026
We earn a commission when you sign up through links on this site. Partner status can affect where a brand appears in our rankings.
Our verdict
Beam Cloud suits developers and small teams that call GPU endpoints in short bursts rather than hold servers idle between requests. The hourly plan starts at $1.83 per GPU hour, with a run cap of 180 seconds and cards spanning H100, H200 and B200.
Key facts
- Best for
- Short, usage-metered GPU inference jobs.
- GPU models
- H100, H200, B200, A100, RTX 4090, RTX 5090 (Source, checked September 26, 2026)
- GPU hour from
- $1.83 (Source, checked September 26, 2026)
- Legal entity
- Smartshare, Inc. (Source, checked September 26, 2026)
- Headquarters
- United States (Source, checked September 26, 2026)
- Founded
- 2021 (Source, checked September 26, 2026)
- Support channels
- Help-centre, Community (Source, checked September 26, 2026)
- Pricing model
- Hourly billing
- Official website
- beam.cloud
Serverless
- Price per GB-second
- $0.0000021 to $0.0000055 (Source, checked September 26, 2026)
- Maximum run time
- 180 (Source, checked September 26, 2026)
- Legal entity
- Smartshare, Inc. (Source, checked September 26, 2026)
- Headquarters
- United States (Source, checked September 26, 2026)
- Founded
- 2021 (Source, checked September 26, 2026)
- Support channels
- Help-centre, Community (Source, checked September 26, 2026)
Pros and cons
Pros
- Endpoint work meters by usage at $0.0000021 to $0.0000055 per GB-second rather than by server time.
- The card line-up spans H100, H200 and B200.
- Each call stops at 180 seconds, which bounds a stuck job.
Cons
- On entry GPU price, Beam Cloud sits 71 of 95 brands.
- Support runs on help centre and community only.
- GPU work bills at hourly rates, so the per GB-second meter at $0.0000021 to $0.0000055 does not reach it.
Who it suits
- Small teams that want H100, H200 and B200 for short bursts rather than a permanent rig.
- Builders comfortable with a 180 second ceiling on every call.
- Anyone who prefers usage metering at $0.0000021 to $0.0000055 to paying for idle capacity.
Who should look elsewhere
- Buyers optimising purely for entry price, since the hourly plan starts at $1.83 against a site median of €0.63; Salad is worth a side-by-side check on NVIDIA RTX 3060, NVIDIA RTX 3070 and NVIDIA RTX 3070 Ti.
- Teams needing runs past 180 seconds, or a card outside H100, H200 and B200; ComputePrices.com tracks H100 SXM, A100 SXM and H200.
- Shoppers who want a live support desk, since support runs on help centre and community; SberCloud lists B300, H100 and A100 for buyers comparing on hardware.
Who runs it
Beam Cloud is operated by Smartshare, Inc., a company headquartered in the United States and founded in 2021. The service it runs is a serverless endpoint for GPU models, sold on hourly billing. Support is limited to help centre and community, so questions get answered in documentation and community threads rather than by a staffed desk.
How Beam Cloud compares
Beam Cloud vs BytePlus8.7 vs 8.7
Beam Cloud vs RunPod8.7 vs 9.5
Beam Cloud vs Netlify8.7 vs 8.6
Beam Cloud vs ScalaHosting8.7 vs 8.3
Beam Cloud vs Hostinger8.7 vs 9.1
Beam Cloud vs Proton Drive8.7 vs 8.3
Beam Cloud vs AIS Cloud8.7 vs 8.1
Frequently asked questions
How much is a GPU hour on Beam Cloud?+
The hourly plan starts at $1.83 per GPU hour, which sits above the site median of €0.63. That is the entry figure rather than a cap, since the card you pick and the length of the call decide what a job actually costs.
How does the per GB-second meter work?+
Serverless endpoints are priced at $0.0000021 to $0.0000055 per GB-second, so a call is charged for the memory and time it uses rather than for a whole server. GPU work sits on a separate hourly plan starting at $1.83.
How long can a single request run?+
A single call runs for up to 180 seconds. Longer jobs have to be split into shorter calls or placed elsewhere, because the endpoint model is built around quick inference metered at $0.0000021 to $0.0000055 per GB-second.
Which GPUs are available?+
The line-up covers H100, H200 and B200, with the hourly plan starting at $1.83. If the card you want is not in that list, ComputePrices.com tracks H100 SXM, A100 SXM and H200, which widens the search before you settle on a rate.
What support do you get?+
The only channels named are help centre and community, so answers come from documentation and from other users. That works for builders who read the docs and ask peers, while teams wanting a staffed desk for production incidents will have to look elsewhere.
Who runs Beam Cloud?+
The service runs under Smartshare, Inc., headquartered in the United States and founded in 2021. Billing is one hourly GPU plan from $1.83 plus endpoint usage metered at $0.0000021 to $0.0000055 per GB-second.
Is Beam Cloud cheap against other GPU hosts?+
On entry GPU price, Beam Cloud ranks 71 of 95 brands, with a site median of €0.63 an hour. Its overall score is 2.8 and the cost part is 2.8.
How does it compare with Salad or SberCloud?+
Salad lists NVIDIA RTX 3060, NVIDIA RTX 3070 and NVIDIA RTX 3070 Ti and SberCloud lists B300, H100 and A100, so card choice is the real difference rather than the billing model. Beam Cloud's own range is H100, H200 and B200, with hourly GPU work starting at $1.83.
Sources
- beam.cloud/pricing checked September 26, 2026
Backs: Price per GB-second
“RAM $0.0000021/sec · per GiB ... CPU and memory · attached to a GPU: RAM $0.0000055/sec · per GiB”
- docs.beam.cloud/v2/endpoint/overview checked September 26, 2026
Backs: Maximum run time
“Endpoints are RESTful APIs, designed for synchronous tasks that can complete in 180 seconds or less.”
- docs.beam.cloud/v2/security/terms-and-conditions checked September 26, 2026
Backs: Headquarters, Legal entity
“Smartshare, Inc. d/b/a Beam 1 Broadway, 14th Floor, Cambridge, MA 02142, United States”
- beam.cloud checked September 26, 2026
Backs: GPU models, Support channels
“H100 80GB VRAM, B200, RTX 4090, H200, RTX 5090, A100”
- beam.cloud/pricing checked September 26, 2026
Backs: GPU hour from, Pricing model
“H100s from $1.83/hr”
- ycombinator.com/companies/beam checked September 26, 2026
Backs: Founded
“Founded:2021”
Compiled from 6 source pages, checked September 26, 2026. We did not open an account.
Not confirmed yet: Billing granularity, Egress fee.