Open Source Twilio Alternatives

LiveKit provides scalable multi-user conferencing built on WebRTC, together with the client SDKs needed to put video, audio and data into an application. It removes the months of work that go into assembling media servers, signalling and connectivity before a first call ever connects.

The server is written in Go on top of the Pion WebRTC implementation and runs as a distributed selective forwarding unit. Deployment is a single binary, a Docker image or Kubernetes, and JWT authentication is built in for production use.

The feature set covers what live calls tend to need once they leave a demo.

• Networking: UDP, TCP and TURN, with distributed multi-region deployments for participants spread across the world.
• Media handling: simulcast, selective subscription, speaker detection and SVC codecs including VP9 and AV1.
• Security: end-to-end encryption, plus moderation APIs for managing participants in a room.
• Egress and Ingress: record or multi-stream rooms and export tracks, and pull in RTMP, WHIP, HLS or OBS Studio sources.
• Agents: a separate SDK for code-first realtime voice agents in Python and Node.js, bundling STT, LLM, TTS, turn detection and telephony.

Client SDKs cover JavaScript and TypeScript among other languages, with declarative UI helpers for React. Live demos include a meeting app, a spatial audio room and livestreaming from OBS Studio, each with published source you can read before committing to the stack.

Telephony bolted onto an application usually arrives as a metered vendor API. Fonoster offers an open source alternative to Twilio, a programmable telecommunications stack connecting telephony services to the internet through a cloud-based utility, for businesses that want call handling on their own terms.

A Voice Application here is a server that controls the flow of a call, written with the Node.js SDK. Verbs such as answer, say, gather, dial, record and hangup compose the logic, speech is synthesized through Google Speech APIs, and a bidirectional stream carries audio both ways. Everything is API first, so originating a call is a handful of lines through the same SDK.

The platform covers the operational surface a phone system needs.

• Multitenancy: separate tenants served from a single deployment.
• PBX features: private branch exchange functionality deploys without manual assembly.
• Access control: OAuth2 and JWT authentication paired with role-based access control.
• Secure endpoints: API endpoints protected with Let's Encrypt certificates.
• Storage: recordings and media held through Amazon S3 support.

A plugin-based command-line tool handles administration, and self-hosting is documented as a Docker deployment. The project describes itself as researching this stack and still frames onboarding through an early access quickstart, so treat it as a platform under active development rather than a settled product. Development happens in the open, with issues and discussion on GitHub and a sizable Discord community around it.

Ethora packages real-time messaging, AI bots and the backend behind them into SDKs you drop into an existing application. The repository itself is a monorepo that gathers every SDK, tool and sample app as a git submodule, so one clone gives you the whole ecosystem to navigate.

Messaging runs over XMPP with a REST API in front of it. A set of hosted services covers the web app, the API, the XMPP gateway and a playground for trying the React chat component and exporting code before writing any of your own. Every SDK also supports pointing at a self-hosted server.

The clients are published separately and tracked against a shared feature matrix.

• React: a JavaScript chat SDK for embedding conversations in any web app.
• React Native: one component covering iOS and Android.
• Android: a native Kotlin SDK.
• iOS: a native Swift SDK.
• WordPress: a plugin that adds a chat widget with no code.
• Bots and agents: an XMPP and LLM bot framework, RAG demos, a site crawler and an MCP server for IDE and agent integration.

Shared capabilities include sending, editing, deleting and replying to messages, reactions, typing and delivery indicators, paginated history, media, one to one and group rooms, unread counters, push notifications, prebuilt UI and an offline cache, with coverage varying by platform. A setup command scaffolds a project, and the matrix is reviewed monthly.

SAMA is a chat server and a messaging protocol written as a deliberate alternative to XMPP, which its authors call the only widely spread standard in this space. It arrives as a server plus working clients rather than a specification to implement yourself.

The server runs on Node.js with uWebSockets, keeps its data in MongoDB and uses Redis for cluster coordination. Nodes find each other through Redis records carrying each node's WebSocket address, establish links with a ping and pong handshake, and retry on a widening delay when a connection breaks. When a node vanishes for good, a surviving node clears the records it left behind, so user and device state does not go stale.

The API is documented one area at a time.

• Users: accounts, authentication and the calls around a user record.
• Conversations: creating and managing conversations between users.
• Messages: sending, delivering and retrieving message history.
• Activities: presence and status for connected devices.
• Address book: contact lists reconciled against server side users.
• Push notifications: delivery to devices that are not currently connected.

The server is GPL 3.0, and clients exist for the web and for Flutter in their own repositories. Local development, full deployment and end to end tests are documented on the project's docs site, and the whole stack can be exercised on a public cloud instance before anything is installed.

Wraps sets up email infrastructure inside your own AWS account with one CLI command, then hands you a TypeScript SDK to send through it. The middleman disappears: you pay AWS directly for sending, and if the project vanished tomorrow the pipeline it built would keep working.

A single init run provisions the whole stack, namespaced under a common prefix and tagged as managed by the CLI, so it stays obvious what belongs to Wraps. Sending afterwards is a few lines of typed code, with visual templates, analytics and a dashboard on top.

That one command creates the pieces an email pipeline needs.

• SES: domain verification with DKIM, SPF and DMARC records handled for you.
• EventBridge and SQS: bounces, opens, clicks and deliveries arrive as real-time events.
• DynamoDB: email event history is retained for a configurable window.
• Lambda: event processing and webhook handling run as functions in your account.
• IAM: least-privilege roles with OIDC support for Vercel and AWS native workloads.

SMS has its own CLI and SDK, and CDK and Pulumi packages cover infrastructure as code. Node 20 or newer plus configured AWS credentials are the only prerequisites. The code is AGPL-3.0 with enterprise features under a commercial license, and the project publishes an agent-readable documentation bundle for people working through Claude Code or Cursor.