Operator · Cambridge, MA
Beam
US serverless GPU and sandbox platform (Python/TypeScript SDK, open-source beta9 engine) billed by the millisecond, with on-demand GPU machines, reserved InfiniBand clusters and bring-your-own-cloud.
GPU types9with a current offer
Current offers4343 readings on file · last 45 days
H100 from$1.83per GPU-hour, on-demand
Entry price$0.44cheapest GPU-hour listed
Countries13 named regions
Trust Score v0.176/100
- Company & track record73
- Infrastructure control100
- Pricing & billing transparency85
- Reliability signals55
- Security & compliance73
- Customer signal—
Scored Sep 29, 2026 from the provider's own pages. Dimensions not yet measured are excluded from the denominator. Criteria ↓
Trust Score · 5 of 6 dimensions measuredThe score, criterion by criterion
Each point comes from something the provider publishes, with the page it was read on and the date. Anything not disclosed earns zero.
Company & track record
weight 15%
73 / 100
| Founding year published | not disclosedbeam.cloud · Sep 29, 2026 | 0 / 4 |
| Headquarters published | Smartshare, Inc.beam.cloud · Sep 29, 2026 | 3 / 3 |
| Legal entity named | Smartshare, Inc.beam.cloud · Sep 29, 2026 | 4 / 4 |
| Funding or ownership disclosed | Backed by Y Combinator (home page); funding amounts not published on pages readbeam.cloud · Sep 29, 2026 | 4 / 4 |
Infrastructure control
weight 15%
100 / 100
| Capacity model published | operates or rents dedicated capacitybeam.cloud · Sep 29, 2026 | 10 / 10 |
| Regions published | 3 regions or countriesbeam.cloud · Sep 29, 2026 | 5 / 5 |
Pricing & billing transparency
weight 20%
85 / 100
| Public prices without an account | yes (read on the pricing page)beam.cloud · Sep 29, 2026 | 5 / 5 |
| Billing granularity published | per secondbeam.cloud · Sep 29, 2026 | 4 / 4 |
| Egress policy published | included (no egress or bandwidth fees on any plan)beam.cloud · Sep 29, 2026 | 4 / 4 |
| Stopped-instance billing published | Serverless: nothing billed when containers are stopped; storage volumes not billed as compute (free up to 1 TB). Default keep-warm after requests is billed: 180 s endpoints, 10 s task queues, 600 s podsbeam.cloud · Sep 29, 2026 | 4 / 4 |
| Public IPv4 policy published | not disclosedbeam.cloud · Sep 29, 2026 | 0 / 3 |
Reliability signals
weight 20%
55 / 100
| Status page | https://status.beam.cloud/beam.cloud · Sep 29, 2026 | 6 / 6 |
| SLA published | not disclosedbeam.cloud · Sep 29, 2026 | 0 / 6 |
| SLA with a numeric uptime commitment | not disclosedbeam.cloud · Sep 29, 2026 | 0 / 3 |
| Uptime history published | Status page shows 100% API uptime over the past 90 daysbeam.cloud · Sep 29, 2026 | 5 / 5 |
| Observed availability (30 days) | not measured yet | — |
Security & compliance
weight 15%
73 / 100
| SOC 2 | Type IIbeam.cloud · Sep 29, 2026 | 8 / 8 |
| ISO 27001 | not disclosedbeam.cloud · Sep 29, 2026 | 0 / 4 |
| HIPAA or GDPR DPA | BAA for Enterprise customers under an Order Formbeam.cloud · Sep 29, 2026 | 3 / 3 |
Customer signal
weight 15%
not measured
| Aggregated public reviews | not collected yet | — |
provisional v0.1: evidence from the provider's own pages; dimensions not yet measured are excluded from the denominatorFull method →
Market report · Oct 25, 2026 · Specialised GPU clouds, hourly rental, pods and inference · #74 of the 200 to audit · prices not read in the reportWhat the market report notes about Beam
Our internal report lists 248 actors with their zone, segment, status and the prices it could read, dated Oct 25, 2026. Its notes are reproduced as written, in French, with their sources.
- Listed as
- Beam
Prices by GPU · 9 GPU types with a current offerWhat Beam charges, against the market
Cheapest current on-demand true cost here, next to the market median for the same GPU. Negative means cheaper than the market.
| GPU | Cheapest here | Market median | vs market | Current offers | Modes | |
|---|---|---|---|---|---|---|
| H100 | $1.83 | $3.17 | −42% | 1 | on-demand | H100 prices |
| A100 80GB | $1.36 | $1.99 | −32% | 1 | on-demand | A100 80GB prices |
| RTX PRO 6000 | $1.09 | $2.14 | −49% | 1 | on-demand | RTX PRO 6000 prices |
| H200 | $2.09 | $4.76 | −56% | 1 | on-demand | H200 prices |
| L40S | $0.76 | $1.26 | −40% | 1 | on-demand | L40S prices |
| B200 | $4.11 | $7.50 | −45% | 1 | on-demand | B200 prices |
| RTX A6000 | $0.54 | $0.54 | −0% | 1 | on-demand | RTX A6000 prices |
| RTX 5090 | $0.72 | $0.69 | +4% | 1 | on-demand | RTX 5090 prices |
| RTX 4090 | $0.44 | $0.44 | −0% | 1 | on-demand | RTX 4090 prices |
All offers · 43 readings on fileEvery Beam offer, with its receipt
Current offers (last 45 days) ranked by true cost, all rental modes, each with its read date. Older readings stay listed below, in grey.
| GPU | Configuration | Mode | Region | Stock | List | True cost | Read | |
|---|---|---|---|---|---|---|---|---|
| RTX 4090 | 1× RTX 4090 | On-demand | — | Stock unknown | $0.44 | $0.44 | 21:35 UTCfresh | Receipt |
| RTX A6000 | 1× RTX A6000 | On-demand | — | Stock unknown | $0.54 | $0.54 | 21:35 UTCfresh | Receipt |
| RTX 5090 | 1× RTX 5090 | On-demand | — | Stock unknown | $0.72 | $0.72 | 21:35 UTCfresh | Receipt |
| L40S | 1× L40S | On-demand | — | Stock unknown | $0.76 | $0.76 | 21:35 UTCfresh | Receipt |
| RTX PRO 6000 | 1× RTX PRO 6000 | On-demand | — | Stock unknown | $1.09 | $1.09 | 21:35 UTCfresh | Receipt |
| A100 80GB | 1× A100 80GB | On-demand | — | Stock unknown | $1.36 | $1.36 | 21:35 UTCfresh | Receipt |
| H100SXM | 1× H100 SXM | On-demand | — | Stock unknown | $1.83 | $1.83 | 21:35 UTCfresh | Receipt |
| H200 | 1× H200 | On-demand | — | Stock unknown | $2.09 | $2.09 | 21:35 UTCfresh | Receipt |
| B200 | 1× B200 | On-demand | — | Stock unknown | $4.11 | $4.11 | 21:35 UTCfresh | Receipt |
9 fresh offers · prices per GPU-hour · true cost includes the add-ons the provider publishesOpen in the comparison tool →
True cost & billing · read Sep 29, 2026What the hourly price does not say
Billing granularity, minimums, what a stopped machine costs, traffic, addresses, deposits and identity checks, as published by the provider.
Billing
- Billing granularity
- per second
- Minimum charge
- not published
- Stopped instance
- Serverless: nothing billed when containers are stopped; storage volumes not billed as compute (free up to 1 TB). Default keep-warm after requests is billed: 180 s endpoints, 10 s task queues, 600 s pods
- Egress
- included (no egress or bandwidth fees on any plan)
- Public IPv4
- not published
- Billing model
- both
- Minimum deposit
- not published
- Free credits
- not published
Account
- Payment
- card · crypto (not published)
- Sign-up
- email, card (valid payment method via Stripe for plans that incur fees)
- Identity check
- not published
- Support
- Slack community, email ([email protected]), live chat (Team plan), private Slack channel (Growth plan)
- Enterprise support
- not published
- SLA
- not published
- Status page
- status.beam.cloud ↗
- Published uptime
- Status page shows 100% API uptime over the past 90 days
Infrastructure
- Regions
3 named regions in 1 countries
United States (default) (United States) · Europe (None) · Asia (None)- Countries
- United States
- Platform
- api, cli, serverless, jupyter, custom_docker_image, persistent_volumes, snapshots, bare_metal · SDK : python, typescript · Terraform : — · Kubernetes : no · interconnexion : InfiniBand on reserved multi-node clusters · clusters : Reserved monthly or yearly multi-node 8-GPU H100/H200/B200/B300 SXM clusters with InfiniBand, contact sales
- Storage
- volume · included up to 1 TB; 0.021 USD/GB-month above 1 TB; snapshots included · Distributed storage volumes
- Migration out
- Code is written against the Beam SDK decorators (endpoint, task_queue); moving off means rewriting to plain containers or self-hosting the open-source beta9 with changed imports. Custom Docker images and volumes are portable; no egress fees.
Security & compliance
- SOC 2
- Type II
- ISO 27001
- not published
- HIPAA
- BAA for Enterprise customers under an Order Form
- GDPR / DPA
- yes (DPA with EU SCCs and UK Addendum applies to every customer)
- Other
- Runs on AWS and Google Cloud; no own data centers
- Legal entity
- Smartshare, Inc.
- Funding
- Backed by Y Combinator (home page); funding amounts not published on pages read
GPUBank market study, read on the provider's own pages · 10 pages read · facts marked "not published" were looked for and not foundPages read ↓
From the study · assessmentBeam, in short
What stood out when we read the documents: strengths, weaknesses, who it suits, and what we could not find.
Strengths
- Millisecond billing on serverless GPUs with no charge for machine start or image pull.
- On-demand machines priced low with CPU/RAM/NVMe included: H100 PCIe from 1.83, H200 from 2.09, B200 from 4.11 USD/hour.
- Open-source engine (beta9) for self-hosting and a BYOC mode (management fee 0.019 USD/vCPU-hour and 0.009 USD/GB RAM-hour).
- No egress or bandwidth fees; storage free up to 1 TB.
- SOC 2 Type II audited; HIPAA BAA for Enterprise.
- Python and TypeScript SDKs, CLI, memory snapshots for sub-second cold starts.
Weaknesses
- Serverless GPU prices exclude CPU and RAM: an RTX 4090 endpoint with 2 cores and 16 GiB costs 1.77 USD/hour per Beam's own example.
- Serverless H100 at 3.4992 USD/GPU-hour plus CPU/RAM; H100 and RTX 5090 serverless listed under 'With committed spend'.
- Multi-GPU per workload by request only (Slack).
- No SLA published; region choice requires contacting Beam; default region United States.
- On-demand prices are 'from' prices; docs example CLI output shows different figures (H100 3.63, A100-80 1.49, RTX4090 0.66 USD/hour).
- Seats and GPU concurrency gated by plan (Developer: 5 GPU containers; Team USD 89/month: 50).
Ideal for
- Python developers deploying inference endpoints, task queues or agent sandboxes that scale to zero
- Teams wanting to burst from their own AWS/GCP credits to a managed GPU pool
Not published
- 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.
- Identity verification policy 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.
Observations
Pricing page shows serverless per-second GPU rates, on-demand 'from' hourly prices, cluster 'Contact us' rows and a cost calculator without an account
Headline 'H100s from $1.83/hr' refers to the on-demand PCIe machine
Serverless GPU rates are shown per second and per hour toggle
Pages read
- beam.cloud/pricing — Serverless, sandbox, on-demand, cluster, storage, BYOC prices; plans; egress
- docs.beam.cloud/v2/environment/gpu — Serverless GPU list, on-demand GPU list, multi-GPU by request, regions
- docs.beam.cloud/v2/resources/pricing-and-billing — What is billed, keep-warm defaults, storage
- docs.beam.cloud/v2/security/security — SOC 2 Type II, HIPAA BAA, DPA, AWS/GCP infrastructure, default US region
- docs.beam.cloud/v2/security/terms-and-conditions — Legal entity, Stripe card payments, prepaid credits, billing in arrears, suspension
- docs.beam.cloud/v2/self-hosting/overview — Open-source beta9
- docs.beam.cloud/v2/reference/cli — CLI, machine list
- status.beam.cloud/ — Status page and 90-day uptime
- beam.cloud/ — Y Combinator backing, SDK languages, positioning
- beam.cloud/about — Company description
Questions about Beam
What GPUs does Beam offer?
Beam has 9 GPU types with a fresh offer today, for example H100 from $1.83, A100 80GB from $1.36, H200 from $2.09, RTX PRO 6000 from $1.09, L40S from $0.76, B200 from $4.11 per GPU-hour on-demand. The full list with prices and market medians is on this page.
How does Beam bill?
Billing granularity: per second. Stopped instance: Serverless: nothing billed when containers are stopped; storage volumes not billed as compute (free up to 1 TB). Default keep-warm after requests is billed: 180 s endpoints, 10 s task queues, 600 s pods
How trustworthy is Beam?
Its Trust Score is 76/100 (v0.1), read from its own documents. Strong on infrastructure control. Weaker on reliability signals. The score measures transparency and published guarantees, not customer satisfaction; every criterion and its source is listed above.
Where are Beam's data centres?
3 named regions in 1 countries: United States.
How do you pay Beam, and are there credits?
Payment by card, crypto. No sign-up credit is published.