🚀 TEN Framework 0.10 — Powering the Next Wave of Voice AI Agents
We’re excited to announce TEN Framework v0.10, our biggest release yet. This version focuses on making it easier than ever to build real-time, customizable, and cross-language voice AI agents — without sacrificing performance.
🌍 What’s New in 0.10
-
Main Extension with Customizable Logic - In v0.10, TEN introduces a streamlined
main
extension that acts as the central pipeline for your agent. It preserves all of TEN’s cross-language and high-performance benefits, while giving you a clean entry point to inject and customize your business logic — without touching the underlying foundation. For details, check out our Main Extension Guide. -
Unified Realtime Protocols (v2.0) - Upgraded ASR, MLLM Realtime, TTS, and LLM protocols. The new TTS protocol brings WebSocket-based TTS providers with lower latency, context-aware synthesis, and better error/metrics handling.
-
Expanded Vendor Ecosystem - Official support for Sonniox, Speechmatics, and Hume.ai, and more, alongside existing providers — giving you more flexibility to choose the best tools for your use case.
-
Official Node.js Support - Node.js now joins Python and C++ as a first-class runtime. You can reuse all existing C++/Python extensions and orchestrate your entire agent in Node.js.
-
TMAN Designer UI Upgrade - A refreshed UI makes designing and managing pipelines faster, simpler, and more intuitive.
🙌 Join Us
TEN Framework is open source and community-driven. Try out v0.10 and tell us what you build — your feedback will shape what comes next.
- ⭐ GitHub
- 📖 Docs
- 🤗 HuggingFace