{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiacl5n6ylwsxbak25pjtnc5s2blpll2ntqsqsm2xg7kej4cpcu6ba",
    "uri": "at://did:plc:dz7fbvkxedbwlm4sroohfpee/app.bsky.feed.post/3mlmbqhydtbo2"
  },
  "coverImage": {
    "$type": "blob",
    "ref": {
      "$link": "bafkreidbxyxwutuuojshe6gdhppibfwovbsrfummyetstfyrspuoenceii"
    },
    "mimeType": "image/jpeg",
    "size": 104410
  },
  "description": "What's new? Thinking Machines previewed its AI for real-time native exchange over audio, video, and text.",
  "path": "/thinking-machines-announced-new-interaction-voice-models/",
  "publishedAt": "2026-05-11T22:04:00.000Z",
  "site": "https://www.testingcatalog.com",
  "tags": [
    "https://t.co/qMlIz1pHDc",
    "pic.twitter.com/16NMI7Ao70",
    "May 11, 2026",
    "pic.twitter.com/et50Lkyy5s",
    "Source",
    "@testingcatalog",
    "@thinkymachines"
  ],
  "textContent": "Thinking Machines has introduced a research preview of its interaction models, a new AI system built to enable real-time, native collaboration across audio, video, and text without relying on external scaffolding. Unlike conventional turn-based models, this system continuously processes and responds to multiple input modalities, supporting features such as simultaneous speech, time-awareness, and real-time dialog management.\n\n> THINKING MACHINES 🔥: Research preview of a new family of realtime voice models have been announced!\n>\n> > Today, we’re announcing a research preview of interaction models: models that handle interaction natively rather than through external scaffolding.\n>\n> > Our research preview… https://t.co/qMlIz1pHDc pic.twitter.com/16NMI7Ao70\n>\n> — 🚨 AI News | TestingCatalog (@testingcatalog) May 11, 2026\n\nThe architecture leverages a multi-stream, micro-turn design that processes and generates outputs in 200ms increments, maintaining ongoing two-way exchange with users. The system is composed of a time-aware interaction model and an asynchronous background model that handles extended reasoning and tool usage, and seamlessly integrates background results into conversations.\n\n> The model can multi-task!\n>\n> Long thinks the model knows everything, but the model actually searched while listening and responding to him so he didn't notice. pic.twitter.com/et50Lkyy5s\n>\n> — Thinking Machines (@thinkymachines) May 11, 2026\n\nThe preview is currently available to a limited group of researchers, with broader access planned for later in the year. The primary audience includes AI researchers, developers, and organizations exploring advanced human-AI collaboration. Early benchmarks show the TML-Interaction-Small model outperforms existing models in both intelligence and interaction quality, demonstrating low latency and robust multimodal responsiveness. Industry observers have noted that this approach sets a new expectation for naturalistic, real-time AI collaboration. Thinking Machines, the company behind this release, is focused on advancing AI models that prioritize collaborative processes, addressing the limitations of existing turn-based interfaces, and keeping humans in the loop during complex tasks.\n\nSource",
  "title": "Thinking Machines announced new Interaction Voice Models",
  "updatedAt": "2026-05-11T22:04:00.732Z"
}