A guide to choosing where to run
Serverless compute
Compare serverless compute destinations by execution limits, triggers, runtime constraints and state handling.
What to check before choosing
Start with what triggers execution and how long it must run. HTTP requests, queues and scheduled tasks have different timeout, concurrency and retry requirements. Verify those limits in the product documentation instead of treating serverless as one execution model.
Check runtime and resource constraints: supported languages, native dependencies, CPU and memory allocation, cold starts and outbound network access. Work that needs an always-on process or durable local files may fit an application host or persistent environment better.
This list includes products classified as serverless compute. It is not a list of every platform with a function feature. Use each profile to inspect evidence, state storage options and billing units; request, duration and resource charges are not interchangeable.
Products in this category
- Serverless · Kata / gVisorOpen core$89/mo
Beam
Serverless GPUs and Sandboxes
Serverless · ContainerOpen core$3/moChutes
Breakthrough Serverless Compute for AI, at Scale.
- Serverless · Virtual machineHosted$20/mo
Deno Deploy
One simple platform for all your JavaScript and TypeScript applications.
Serverless · Virtual machineHosted$29/moKoyeb
High-performance Infrastructure for APIs, Inference, and Databases — deploy intensive applications across GPUs, CPUs, and Accelerators in minutes
- Serverless · WebAssemblyOpen source tool—
Kyushu
A Self-Hostable Wasm Sandbox for JavaScript Workers — run a Cloudflare Workers-style handler on a VPS or anywhere, without Node.js, Bun, or even Docker.
- Serverless · Kata / gVisorHosted$250/mo
Modal
The platform for production AI — cloud infrastructure for teams that develop, train, and serve AI applications at scale.
- Serverless · ContainerOpen source tool—
NVCF (NVIDIA Cloud Functions)
Self-managed GPU functions, streaming and batch workloads on Kubernetes
- Serverless · Not disclosedHosted—
Recurse
A serverless harness for agents to create custom agents and deploy them as tools, MCPs, or bots.
- Serverless · ContainerOpen core$20/mo
Rivet
The Orchestrator for Agentic Workloads