Beta9
Ultrafast serverless GPU inference, sandboxes, and background jobs
Last verified:
What is Beta9?
Beam Cloud (Beta9) is an AI-native cloud platform designed for developers who need to run serverless GPU workloads, sandboxes, task queues, and custom model inference with ultrafast boot times and instant autoscaling. The platform lets you run sandboxes for AI agents, durable task queues for processing large amounts of data, and custom model inference on GPU or CPU. It is 100% open source and can run on Beam's cloud or your own infrastructure, giving you flexibility to deploy anywhere.
Key features include secure execution of LLM-generated code in isolated sandboxes, launching containers in under 1 second, deploying existing Docker images as REST APIs instantly, and automatically scaling jobs to thousands of containers. The platform supports training and fine-tuning from SLMs to LLMs and diffusion models, audio processing pipelines, Streamlit and Gradio UIs, and web scraping with Chromium instances. You can switch hardware in seconds by changing one line of Python, do easy local debugging with production-matching configurations, run multiple workers per container, run Docker-in-Docker, and deploy from GitHub Actions.
Beam Cloud is for machine learning engineers, AI developers, data scientists, and teams building AI applications who want to avoid managing infrastructure. It's ideal for hosting custom models, running untrusted code securely, building AI agents, deploying APIs quickly, and scaling workloads without timeout limits. Thousands of developers use Beam to build applications without thinking about cloud infrastructure, with companies like Jamie, Frase, and Happy Accidents trusting the platform.
The platform offers a Pythonic SDK for deploying and scaling AI applications with zero infrastructure overhead. You only pay for the compute you use, billed by the millisecond, with no charges for cold starts or container image loading. The developer experience is praised as top-notch, with responsive support and an active Slack community for quick help when stuck.
Beta9 pricing
Pricing model: Freemium
Free tier: $0 per month plus usage, includes $30 monthly credits, unlimited apps, unlimited storage volumes, unlimited custom images, 5 GPU containers concurrency, 30 CPU containers concurrency, unlimited API requests, 30 days log retention, 1 seat, community support, and all core features (Cloud Storage Volumes, Secrets Manager, Deployment Logs, Export Your Code, Keep Warm, Autoscaling). Basic plan: $89 per month plus usage, includes $30 monthly credits, same unlimited features, 50 GPU containers concurrency, 1000 CPU containers concurrency, 3 seats included ($25 per additional seat), live chat support. Enterprise: Contact Us for custom pricing with custom GPU/CPU concurrency, unlimited containers, unlimited seats, 1 year log retention, private Slack channel, and private support.
Beta9 pros
- Ultrafast boot times with containers launching in under 1 second
- Instant autoscaling to thousands of containers automatically
- 100% open source platform you can run on their cloud or yours
- Serverless GPU workloads with fast cold starts on bare-metal servers
- Pay only for compute used, billed by millisecond of usage
- No charges for cold start or container image loading time
- Secure sandboxed execution for LLM-generated code in non-root containers
- Switch hardware in seconds by changing one line of Python
- Easy local debugging with exact production configuration matching
- Deploy any existing Docker image as REST API instantly
- Long-running functions with no timeouts and scheduled jobs support
- Multiple workers per container for vertical scaling
- Docker-in-Docker support in containers
- GitHub Actions deployment for automatic CI/CD integration
- $30 free monthly credits refreshed every month for all plans
- Self-hosted option keeping all data in your own VPC
- Python SDK with simple 6-line code deployment for GPU apps
Beta9 cons
- Free tier limited to 5 GPU containers concurrent
- Free tier limited to 30 CPU containers concurrent
- Basic plan ($89/month) limited to 3 seats
- Log retention only 30 days on free and basic plans
- Community support only on free tier without live chat
- Higher-end GPUs like B200SXM6 start at $3.93/hour
- Requires Python virtual environment setup for SDK installation
- API token configuration needed before first use
- Enterprise features require contacting sales for custom pricing
- Limited to 1 seat on free plan without paying for additional seats
Frequently asked questions about Beta9
How much does storage cost?
Storage is included free of charge with all Beam Cloud plans. You get unlimited cloud storage volumes on the free tier, basic plan, and enterprise plan without any additional cost beyond your usage fees.
Am I billed for cold start?
No, you are not billed for cold start. Beam only charges for the time to load your application code. They do not charge for the time to spin up a server or load your container image, which means you pay only for actual execution time.
I don't want serverless. Can I use Beam while keeping the server running 24/7?
Yes. By default, apps will spin down automatically after each request, but you can control how long your apps remain active. Many customers with latency-sensitive workloads choose to keep their servers running 24/7 on Beam for consistent performance.
I have an existing Docker image. Can I use it on Beam?
Yes. You can import any base image from a third-party image registry onto Beam. The platform lets you instantly deploy any existing Docker image as a REST API without needing to rebuild or modify your container.
How secure is Beam?
Workloads are isolated from one another and run in non-root containers for security. Beam also offers a self-hosted product that runs entirely in your own environment, ensuring that no data leaves your VPC for maximum security and compliance.
What GPUs are available on Beam Cloud?
Beam offers a wide fleet of GPUs including B200SXM5 ($3.93/hr), H200SXM5 ($1.99/hr), H100PCIE ($1.74/hr), A100 80GBSXM4 ($1.30/hr), L40SPCIE ($0.72/hr), RTX PRO 6000PCIE ($1.04/hr), A6000PCIE ($0.51/hr), RTX 5090PCIE ($0.68/hr), and RTX 4090PCIE ($0.42/hr) for on-demand pricing, plus serverless options for RTX 4090 and A10G billed per second.
How do I get started with Beam Cloud?
Create an account on the Beam platform at platform.beam.cloud to get $30 of free credit refreshed monthly. Activate a Python virtual environment, then retrieve your API token from the dashboard API Keys page and install the SDK with pip install beam-client and beam configure default --token [TOKEN]. You'll get 15 hours of free credit when you sign up.
Can Beam scale to thousands of containers?
Yes, Beam automatically scales jobs to thousands of containers. The platform deploys task queues, scheduled jobs, and long-running functions with instant autoscaling, and you can scale vertically by running multiple workers on the same container.
What types of workloads can I run on Beam?
You can run serverless workloads on CPUs and GPUs, securely execute LLM-generated code in sandboxes, train and fine-tune models from SLMs to LLMs and diffusion models, deploy custom model inference, build audio processing pipelines, run Streamlit and Gradio UIs, perform web scraping with Chromium instances, and execute long-running functions with no timeouts.
What support options are available on Beam?
Free tier includes community support only. The $89/month basic plan adds live chat support. Enterprise customers get access to a private Slack channel and private support. The Slack community is noted as a game changer with quick responses when you get stuck.