Home / Directory / AI infrastructure / Inference & model hosting / Beam
Beam
Serverless GPU and sandbox cloud from a Y Combinator team that runs inference, agents, and task queues with fast cold starts on its cloud or the buyer's own.
AI startups and developers serving open models, running agent sandboxes, or batch jobs who want Python-first deploys and pay-per-use GPUs.
Large enterprises that need a long vendor track record and dedicated account teams, or workloads locked to one hyperscaler's managed services.
Verdict
Beam is an AI compute company founded in 2021 by Eli Mernit and Luke Lombardi, first under the name Slai, and part of Y Combinator's Winter 2022 batch. Developers deploy Python functions as GPU endpoints, sandboxes, persistent VMs, and task queues with one SDK, on Beam's cloud or on their own AWS, GCP, or bare metal, with RTX 4090s from $0.69 an hour. It matters because idle GPUs are expensive, and Beam's memory snapshots make scale-to-zero practical without long cold starts.
Score Breakdown
How Beam scores in the categories that matter to its buyers.
Pricing
Beam bills by the millisecond while code runs. GPUs, CPU, and memory are priced separately on serverless.
| Plan | Price | What stands out |
|---|---|---|
| Serverless RTX 4090 | $0.69 / hour | GPU only; CPU and RAM metered separately |
| Serverless H100 | $3.50 / hour | GPU only; CPU and RAM metered separately |
| Sandbox, 1 core / 8 GiB | $0.319 / hour | Billed per second while running |
| Bring your own cloud | Management fee | Example: $0.44 / hour in Beam fees on an AWS g5.2xlarge |
The Field at a Glance
How Beam compares with other Inference & model hosting vendors we reviewed, by Overall Score and relative typical engagement cost.
Beam scores 7.8 in Inference & model hosting, under Modal (8.4) and Replicate (8.0), level with Cerebrium (7.8), and ahead of Together AI (7.3). Relative cost lands lower for this subcategory.
Use-case matrix
| Use case | Fit | Notes |
|---|---|---|
| Serverless model inference | Strong | Sub-second cold starts from memory snapshots. |
| Agent sandboxes | Strong | Snapshot, branch, and restore into many parallel runs. |
| Using existing cloud credits | Strong | BYOC runs workloads in AWS, GCP, or Azure. |
| Batch and task queues | Strong | Retries, callbacks, and scheduled jobs built in. |
| Multi-node training | Mixed | Reserved InfiniBand clusters are offered separately. |
| Large enterprise procurement | Weak | A young company with a developer-led sales motion. |
Who it’s for
Good fit
- AI startups serving open models
- Teams running agent sandboxes
- Developers with cloud credits to use
Poor fit
- Enterprises needing long vendor history
- Teams standardized on one hyperscaler
- Always-on workloads better suited to reserved hardware
Review Excerpts
Below are excerpts from public reviews. Our team scoured public reviews, forums, and chat rooms to get a balanced view of customers' experience with this company. Paid reviews and pay-for-play sites such as Clutch were excluded.
“Frase is running language models exclusively on Beam and it was surprisingly easy to migrate, less maintenance, and is saving us money because unlike Google and other cloud providers, Beam is able to provide us with an on-demand solution that scales immediately with our traffic, and we don’t need to worry about any of the clunky tooling around GPUs.”
“Beam is powering hands-down the best developer experience to run models on GPUs easily at scale. Best decision on the infra side for us this year so far.”
“Time is the biggest thing Beam has helped us with. I went from spending 6 hours developing an API to pressing a button and deploying instantly”
Serverless GPU rates cover the GPU only; CPU and memory are metered on top, so an RTX 4090 endpoint with 4 vCPU and 16 GiB costs $1.77 an hour rather than the $0.69 headline rate.
Company Facts
- Founded
- 2021 (Y Combinator)
- Headquarters
- New York City, NY (Y Combinator)
- Employees
- About 5 (Y Combinator)
- Accelerator
- Y Combinator, Winter 2022 batch (Y Combinator)
- Website
- beam.cloud
Key Leadership
Eli Mernit
Co-founder and CEO
Methodology
This page is an independent evaluation of Beam for buyers comparing options in inference & model hosting. AI Industry Reviews accepts no sponsorships, advertising, or pay-for-placement fees. Beam did not pay for this review.
What we scored
The headline number is an Overall Score on a 0-10 scale. Eight criteria sit under it in two groups.
- Buyer outcomes (for inference & model hosting)
- Cold start speed
- Developer experience
- Price per GPU hour
- Enterprise track record
- Company & commercial
- Innovation & product leadership
- Project management & communication
- Pricing
- Contract fairness
Pricing measures whether the price looks fair for the value delivered, including packaging and renewal friction that show up in real buying cycles.
Score Composition
| Input | Weight | What it covers |
|---|---|---|
| Reviews | 40% | A proprietary read of what practitioners say about likes, complaints, and day-to-day use, including public review sites, forums, and private chat rooms. Paid reviews and pay-for-play sites such as Clutch are out of scope. |
| Product | 35% | Hands-on look at screens and workflows. |
| Pricing | 15% | Whether the price looks fair for what you get. |
| Docs & training | 10% | Docs, tutorials, and training material. |
How we balanced the evidence
The Overall Score is the simple average of the eight criteria. Recommendation language follows that score and the fit pattern described above.
Scope
Beam is graded here as inference & model hosting. Criteria scores can move as more review volume and product checks are added.