LiveKit Agents 1.8.11 adds Responses API and new speech providers
Updated LiveKit Agents
Summary
LiveKit Agents 1.8.11 (November 5, 2026) adds inference.responses.LLM, Zoom/Kalpa/Rime/Fish Audio plugins, and moves Respeecher WebSocket keys into headers.
- Tool
- LiveKit Agents
- Category
- Voice agents
- Date
- Open source
- Yes
- License
- Apache-2.0
- Pricing
- Not stated in the sources we cited.
- Sources
- GitHub release livekit-agents@1.8.11
- PR 8210: inference.responses.LLM
- PR 8211: WebSocket transport for Responses
- PR 7708: Zoom AI STT
- PR 8017: Kalpa TTS
- PR 8013: Rime STT
- PR 7856: Rime Prism realtime
- PR 8357: Fish Audio realtime STT
- PR 8286: Respeecher auth headers
- PR 8105: OpenAI gpt-realtime-translate
- PR 8106: Gemini 3.5 Live Translate
LiveKit Agents 1.8.11 landed on November 5, 2026. The release notes center on three themes: OpenAI's Responses API through LiveKit Inference, a batch of new STT and TTS providers, and a Respeecher security change that keeps API keys out of WebSocket URLs.
Release facts
| Fact | Detail |
|---|---|
| Product | LiveKit Agents, repository livekit/agents |
| Version | livekit-agents@1.8.11, after 1.8.10 |
| Release date | November 5, 2026 |
| Headline API | inference.responses.LLM(...) with optional WebSocket transport |
| License | Apache-2.0 |
Responses API on LiveKit Inference
PR 8210 and PR 8211 add inference.responses.LLM. You pass a LiveKit Inference model id such as "openai/gpt-4.1". REST is the default; use_websocket=True reuses the connection across turns. Structured output and web search are supported through the same path. Related follow-ups add update_options, response_format, and a reasoning= knob. Inference Responses now defaults each known model to its lowest reasoning effort; set reasoning=None to keep the provider default (PR 8359 in the release changelog).
New speech providers
| Provider | What shipped | PR |
|---|---|---|
| Zoom AI | Zoom Scribe STT with live streaming and per-utterance recognition | 7708 |
| Kalpa | kalpa-tts-1 streaming TTS in English and Hindi | 8017 |
| Rime | STT plus Prism speech-to-speech on the standard realtime interface (Python 3.11+; STT needs the stt extra) | 8013, 7856 |
| Fish Audio | transcribe-stream-1 realtime STT with word timings and recognition context | 8357 |
| OpenAI | gpt-realtime-translate as a beta TranslationModel, with gpt-realtime-whisper input transcription | 8105 |
gemini-3.5-live-translate-preview through Gemini API and Vertex AI | 8106 | |
| Respeecher | Public Swedish, Brazilian Portuguese, German and Arabic TTS models | 8286 |
Provider billing and availability stay separate from installing the framework.
Security and behavior changes
- Respeecher auth: WebSocket API keys move from request URLs into handshake headers (PR 8286).
- STT turn detection: a late final after speech resumes joins the resumed turn and waits for STT's next end-of-speech event.
- Uninterruptible speech: user transcripts are retained without interrupting playback or generating a reply.
- The release notes also list realtime history and reconnection fixes, task-handoff fixes, STT timing and transcript sync, a Python 3.14 VAD buffer copy, and multichannel TTS fallback.
Questions
Is 1.8.11 a new LiveKit Cloud product? No. It is a release of the open-source agents framework. Inference credentials and provider plugins are configured in your agent code.
Do I need WebSockets for Responses?
No. REST is the default. Set use_websocket=True when you want connection reuse across turns.
What broke for Mistral users in 1.8.10? That migration landed in 1.8.10. This release does not reverse it; check that note if you still use Mistral Chat Completions.
For a side-by-side with Daily's framework, see Pipecat vs LiveKit Agents. More tools are in the voice agents category.
Alternatives
| Tool | License | Pricing note |
|---|---|---|
LiveKit AgentsThis launchOpen-source Apache-2.0 framework for realtime voice AI agents in Python and Node.js. Version 1.8.11 (November 5, 2026) adds inference.responses.LLM and new STT/TTS plugins. | Apache-2.0 | The framework is Apache-2.0. LiveKit Cloud and third-party speech/LLM providers bill separately; this page does not quote those prices. |
PipecatOpen-source BSD-2-Clause Python framework from Daily and the community for real-time voice and multimodal agents, built from composable pipelines. | BSD-2-Clause | Not stated |
More from LiveKit Agents
LiveKit Agents 1.8.10 adds recording controls and voice integrations
LiveKit AgentsBitrate settings and new speech plugins arrive with Mistral migration requirements.
Category: Voice agents
Pricing: Not stated
LiveKit Agents 1.8.9 adds decisions through Inference
LiveKit AgentsA feature-flagged DecisionModel batches named decisions through LiveKit Inference; silent tool progress and new speech providers also arrive.
Category: Voice agents
Pricing: Not stated
LiveKit Agents 1.8.6 adds Sarvam TTS and Nabrah STT
LiveKit AgentsLiveKit Agents 1.8.6, released October 9, 2026, adds Sarvam, Nabrah, Speechify and Cerebras updates plus fixes for voice and IPC.
Category: Voice agents
Pricing: Open source
Questions
What is LiveKit Agents?
The documentation says the framework lets you add Python or Node.js programs to LiveKit rooms as realtime participants. The repository describes it as a framework for building realtime voice AI agents.
Is LiveKit Agents open source?
Yes. The GitHub API reports Apache-2.0 for livekit/agents.
What is the latest release covered here?
livekit-agents@1.8.11, published November 5, 2026. It adds inference.responses.LLM (REST by default, optional WebSocket), Zoom AI STT, Kalpa TTS, Rime STT and Prism, Fish Audio realtime STT, OpenAI and Gemini live translate models, and moves Respeecher WebSocket API keys into handshake headers.
Where can agents be deployed?
The documentation says agents can be deployed to LiveKit Cloud or any custom environment.
Filed on Agent Radar. Claims in this note are limited to the sources above.