Yes. The framework is released under the Apache License 2.0 with additional restrictions, while components in the packages directory are under plain Apache 2.0.

What do I need to run the examples locally?

Docker/Docker Compose, Node.js v18, at least 2 CPU cores and 4 GB RAM, plus Agora, OpenAI, Deepgram and ElevenLabs keys.

Can I run it without Docker?

Yes, agent examples can run in GitHub Codespaces, and a beta transcriber app can run from TEN Manager without Docker.

How do I deploy an agent?

Build a release Docker image, or run the backend on a container platform and deploy the frontend to Vercel or Netlify.

TEN Agent

Meet TEN, the World's First Truly Real-time Multimodal Agent Framework for Creating Next-Gen AI Agents

Category:

Builders

Maker:

不写

Use Cases

SaaSDevelopment

⚡️ Tags detected by AI

5 visitors went to TEN Agent from LandinTools in the last 30 days.

Your product? Claim it to see where they come from

About TEN Agent

TEN is an open-source framework for building real-time multimodal conversational AI agents, with a focus on voice. The repository includes the TEN Framework itself along with Agent Examples, a voice activity detector (VAD), Turn Detection for full-duplex dialogue, and the TEN Portal for documentation.

The agent examples cover a multi-purpose voice assistant supporting RTC and WebSocket connections, a doodle board that turns spoken or typed prompts into sketches, real-time speaker diarization, lip sync avatars (including Trulience, HeyGen and Tavus), SIP phone calls, transcription, and an example running on the Espressif ESP32-S3 Korvo V3 board. Developers configure STT, LLM and TTS extensions through the TMAN Designer visual interface or by editing property.json.

The quick start runs in Docker or GitHub Codespaces and requires Agora, OpenAI, Deepgram and ElevenLabs keys. Finished agents can be deployed as Docker images, or split so the backend runs on a container platform and the frontend on Vercel or Netlify. It is aimed at developers building voice and multimodal AI applications.

Best for

Developers building real-time voice AI agents
Teams prototyping AI avatars or phone call agents
Hardware makers adding LLM voice to devices

Pros

Open source with Apache 2.0-based licensing
Many ready-made agent examples, from voice assistants to SIP calls
Visual TMAN Designer for configuring STT, LLM and TTS extensions
Deploys via Docker or split across Vercel/Netlify and container hosts

Cons

Quick start requires Agora, OpenAI, Deepgram and ElevenLabs API keys
Setup requires Docker and developer experience
License includes additional restrictions beyond Apache 2.0

TEN Agent FAQ

Is TEN open source?: Yes. The framework is released under the Apache License 2.0 with additional restrictions, while components in the packages directory are under plain Apache 2.0.
What do I need to run the examples locally?: Docker/Docker Compose, Node.js v18, at least 2 CPU cores and 4 GB RAM, plus Agora, OpenAI, Deepgram and ElevenLabs keys.
Can I run it without Docker?: Yes, agent examples can run in GitHub Codespaces, and a beta transcriber app can run from TEN Manager without Docker.
How do I deploy an agent?: Build a release Docker image, or run the backend on a container platform and deploy the frontend to Vercel or Netlify.