Offering overview
Decorate a Python function and ship it as an autoscaling GPU endpoint: that is the developer experience this Y Combinator-backed company sells. Memory snapshots restore containers up to 35x faster than a conventional cold boot, workloads scale to zero and burst to thousands of containers, and a CLI plus GitHub Actions integration cover deployment. Three products share the runtime: Inference (OpenAI-compatible LLM serving on serverless GPUs), Task Queues (durable background jobs with retries, callbacks and schedules) and Sandboxes (stateful code-execution environments that can be snapshotted and forked, including for RL rollouts). Persistent volumes, filesystem snapshots and Docker-in-Docker are included. GPUs range from RTX 4090 and RTX 5090 through A6000, L40S, RTX PRO 6000, A100, H100, H200 and B200, with multi-node H100/H200/B200/B300 clusters by request. On-demand machines start at $0.44 per hour for an RTX 4090; serverless 4090 time is $0.000192 per second (about $0.69/hr). The Developer plan is $0/month plus usage, Team is $89/month, and BYOC deployments pay a small per-vCPU and per-GB platform fee. The company lists a Cambridge, Massachusetts address.
- Management
- Not stated
- Infrastructure
- Not stated
- Pricing model
- Not stated
- Support
- Not stated
- SLA
- Not stated
- Best for
- Model Inference, AI Agents, AI Applications, Batch Processing, Fine-Tuning, Model Training
Technical specifications
- Containers
- Unknown
- Kubernetes
- Unknown
- Root access
- Unknown
- API access
- Unknown
- Autoscaling
- Unknown
- Persistent processes
- Unknown
AI capabilities
- Model serving
- Not stated
- AI frameworks
- Not stated
- Inference endpoints
- Unknown
- Fine-tuning
- Unknown
- Vector database
- Unknown
- Model catalog
- Unknown
- Monitoring
- Not stated
- Evaluation
- Not stated
Security & compliance
No controls are stated.
Geography
Provider headquarters are separate from offering infrastructure and availability.
Infrastructure regions
Global
Data-residency jurisdictions
Not stated
Service availability
Not stated
Support coverage
Not stated
Available plans
Pricing can change. Promotional prices are shown separately from standard prices.
| Plan | Price | Compute | Storage | GPU | Price checked | Actions |
|---|---|---|---|---|---|---|
| Beam Serverless GPU CloudNot stated | $0.44Usage Only | Unknown — RAM | Unknown | No | 7 October 2026 |
Questions & answers
- How do developers deploy workloads on Beam?
- Developers decorate a Python function to deploy it as an autoscaling GPU endpoint. Beam also provides a CLI and GitHub Actions integration for deployment.
- What workloads does Beam support?
- Beam supports OpenAI-compatible LLM inference, durable background task queues and stateful code-execution sandboxes. Task queues include retries, callbacks and schedules.
- Can Beam run in a customer's own cloud?
- Beam can run on its own cloud or in a customer's AWS, GCP or Azure environment.
- What can Beam Sandboxes do?
- Beam Sandboxes provide stateful code-execution environments that can be snapshotted and forked, including for reinforcement learning rollouts. The runtime also includes persistent volumes, filesystem snapshots and Docker-in-Docker.
Data sources & provenance
Where this information came from, how it was checked and when it was last looked at. “Last updated” records a change to the record — it does not mean the record was verified.
- Data quality
- Data quality: Unverified
- Primary source
- beam.cloud
- Sources cited
- 2
- Last reviewed
- 7 October 2026
- Last updated
- 18 October 2026