The open-source alternatives to ElevenLabs Agents are Pipecat, LiveKit Agents and TEN Framework. ElevenLabs sells the agent platform and the voices together; the open-source split is a framework that connects any speech-to-text, language and text-to-speech model, including ElevenLabs' own voices if you keep them. Pipecat (BSD-2-Clause) lists the most providers and includes local options like Whisper, Kokoro and Piper. LiveKit Agents (Apache-2.0) runs the whole stack on your servers. TEN is source available under Agora's conditions. A team switches when it wants to swap any model in the pipeline without changing platform.
ElevenLabs Agents, in the vendor's own words: "conversational agents that feel human" which you "configure, deploy and monitor" across voice, chat and email, by its own description. elevenlabs.io.
| Project | Openness | Licence | Runs as | Platforms | Self-host | Hosted option | Last release | Stars |
|---|---|---|---|---|---|---|---|---|
| Pipecat | OSI | BSD-2-Clause | pip, CLI | Python 3.11 or later server; client SDKs for JavaScript, React, React Native, Swift, Kotlin, C++ and ESP32 | Yes | Yes | Sep 12, 2026 | 15,594 |
| LiveKit Agents | OSI | Apache-2.0 | pip (Python), npm (the separate AgentsJS repository) | Python server; client SDKs for browsers, Swift, Android, Flutter, React Native, Rust, Node.js, Unity, ESP32 and C++ | Yes | Yes | Sep 15, 2026 | 14,230 |
| TEN Framework | Source available | See LICENSE | Docker Compose, GitHub Codespaces | Docker; Python, Go and Node.js inside; an ESP32-S3 board example | Yes | No | Jul 31, 2026 | 11,130 |
Licence, release and star figures read from GitHub on September 16, 2026. Openness labels: The code, and the weights where the project is a model, carry a licence on the OSI approved list. The code is public, and the licence limits what you may do with it. Read the LICENSE file before commercial use. The model weights can be downloaded and run. The training data or code is closed, or the weights carry a restriction the code does not.
A Python framework under BSD-2-Clause, maintained by Daily, for real-time voice and multimodal agents, with the widest provider list on this site and telephony serialisers for Twilio, Telnyx, Plivo, Vonage, Exotel and Genesys.
pip, CLI · Python 3.11 or later server; client SDKs for JavaScript, React, React Native, Swift, Kotlin, C++ and ESP32 · Runs on your own machines · Hosted version available from the maintainers
An Apache-2.0 framework for real-time voice agents that connects any speech, language and voice provider, with semantic turn detection, native MCP tool support, a test framework, and an open-source media and SIP stack it can run on entirely.
pip (Python), npm (the separate AgentsJS repository) · Python server; client SDKs for browsers, Swift, Android, Flutter, React Native, Rust, Node.js, Unity, ESP32 and C++ · Runs on your own machines · Hosted version available from the maintainers
A real-time conversational AI framework from Agora with a visual designer, a SIP extension and turn detection, released under Apache-2.0 plus conditions that forbid competing deployments, which makes it source available rather than open source.
Docker Compose, GitHub Codespaces · Docker; Python, Go and Node.js inside; an ESP32-S3 board example · Runs on your own machines
Pipecat (BSD-2-Clause) if you want the widest provider list, including ElevenLabs itself for voices and local models like Whisper, Ollama, Kokoro and Piper. LiveKit Agents (Apache-2.0) if you want the media server, the SIP service and the framework all open source on your own machines, with semantic turn detection. Both are Python; LiveKit also has a Node.js version.
Yes. Pipecat lists ElevenLabs among its text-to-speech providers, and TEN's quick start requires an ElevenLabs key for speech. LiveKit Agents describes a mix-and-match ecosystem of speech, language and voice providers. The framework is open; the voice can stay with the vendor you prefer.
LiveKit Agents states that the entire stack, including the LiveKit media server, can run on your own servers. Pipecat runs locally and offers eight transports, one of which is a local transport. TEN's documented quick start depends on Agora's real-time service; its README describes self-hosting the TEN backend but not replacing that service.
LiveKit Agents has native MCP support, described as integrating tools from MCP servers in one line. Pipecat lists function calling across its language model providers and a skills package for Claude Code. TEN describes extensions rather than MCP in its README.
Pipecat lists Whisper and Moonshine for speech to text, Ollama for the language model, and Kokoro, Piper and XTTS for text to speech, all of which run on your hardware. LiveKit Agents does not enumerate local models in its README. TEN's quick start uses hosted providers.
Running one of these inside your own environment, with your own data and your own security rules, is the kind of work Reveneau does. Read how a forward deployed engagement works.