AI Trends

Trending story

LiveKit Turn Detector v1.1 Is Now Available

1 reports1 story updatesUpdated Nov 20 08:00

Understand this story

AI overview

LiveKit Turn Detector v1.1 launched on November 20, 2026, and is available for agents hosted on LiveKit Cloud as well as self-hosted agents using LiveKit Inference. According to the LiveKit Product & Engineering Blog, at the same false-interruption rate, it reduces English endpointing latency by 15% to 34% compared with v1, and by 17% to 19% for the other 13 languages. This version reads both the user's speech and the agent's previous text message; the open-weight v1.1-mini version can run on CPUs and supports the same 14 languages. There are currently no other evidence-backed subsequent developments.

AI-generated from public reports

Latest development · Report

LiveKit Releases Turn Detector v1.1, Reducing Endpoint Detection Latency by 15–34%

Developments

1 confirmed developments

  1. LiveKit Turn Detector v1.1 is now available

    Source:English text is being prepared.

Reports and discussion timeline

Browse reports and discussion by time. Their evidence levels remain distinct.

Nov 20

  1. LiveKit Product and Engineering BlogFirst-party

    LiveKit Releases Turn Detector v1.1, Reducing Endpoint Detection Latency by 15–34%

    LiveKit has released Turn Detector v1.1, reducing English endpoint detection latency by 15–34% compared with v1 at the same false interruption rate, and by 17–19% across 13 other languages. v1.1 reads both the user’s speech and the agent’s previous text message to improve recognition of short responses and pauses by non-native speakers. The open-weight v1.1-mini version can run on a CPU and supports the same 14 languages.

Related stories

  1. Related storiesMustafa Suleyman launches MAI Cyber 2 Flash and MAI Code 2
  2. Related storiesTencent Hunyuan open-sources Chinese-English simultaneous interpretation model Hy-SimulMT
  3. Related storiesStepFun updates Step Code to v0.1.4
  4. Related storiesEnglish translation is being prepared.