Rivet

A fast, high-density, scalable orchestrator for every workload an agent needs.

Rivet runs agents, workflows, sandboxes, and generated apps as Actors: long-lived processes with durable state, scheduled by one control plane and hosted in worker processes you already run. Every Actor type on this site is built on the same primitive.

Why Actors

Run thousands of workloads per server. An Actor lives inside a worker process you already run, not in its own container or VM. A running Actor adds about 72 KB of memory (RSS increase per Actor, measured on FoundationDB with the Rust rivetkit SDK); a server that held hundreds of pods holds hundreds of thousands of Actors.

Start in milliseconds, not minutes. A cold start includes scheduling the Actor, loading its state, and serving the first request: 12.3 ms p50 (27.7 ms p99) for an HTTP round trip waking a hibernated Actor, measured on FoundationDB with the Rust rivetkit SDK using an experimental Rivet feature that will be enabled by default in an upcoming update, against roughly 6 s for a Kubernetes pod. No image pull, no container boot.

Built for scale. Actors are scheduled independently, so adding capacity means adding machines. The same control plane runs a thousand Actors or a billion.

Hibernates when idle, wakes on demand. An idle Actor writes its state and unloads. The next request brings it back in milliseconds with nothing lost, and a hibernating Actor holds no memory.

Durable state for every workload. Every Actor gets a SQLite database and a POSIX filesystem, tiered to S3. Idle Actors cost nothing, so millions can sit parked with their state intact.

What the control plane provides

• Fault tolerance by design. A worker going down reschedules its Actors elsewhere with their state intact. No replay harness or checkpointing of your own.
• Multi-region. Place Actors near the users and data they serve, and route requests to wherever each one currently lives. See edge placement.
• HTTP and WebSocket networking. Address an Actor directly over HTTP or hold a live WebSocket to it. No queue or broker in between.
• End-to-end type safety. Actor definitions generate their own client types, so a signature change breaks the build rather than production.
• React SDK. First-party hooks that subscribe a component to an Actor’s state and keep it live as the Actor updates.
• OpenTelemetry. Traces, metrics, and structured logs emitted in OTel format, into the collector you already run.
• Cron and scheduling. Wake an Actor on a schedule or at a timestamp it sets for itself, without a separate scheduler.
• Actor-to-Actor calls. Actors address each other by key and call across the cluster as if the other one were local.
• Single Rust binary. The control plane ships as one static binary. One control plane behind a load balancer, speaking plain HTTP inside your VPC: no CRDs, no operator, no service mesh.
• Open source. Apache 2.0 and self-hostable in full. Rivet Cloud runs the same control plane you can run yourself.

Works with the tools you already use

Typed SDKs for TypeScript, Node.js, Bun, and Rust on the backend, React hooks on the frontend, and an Effect SDK. Start from the quickstarts, or connect your AI client with Rivet MCP and the Rivet skill.

Where to go next

• Guides: worked examples such as an AI agent, a chat room, a multiplayer game, and a per-tenant database.
• Integrations: frameworks and libraries that run on Rivet Actors.
• Deploy: run the control plane on Kubernetes, Docker, or a VM, or deploy workers to a serverless platform.
• Registry: every Actor type, first-party and community.