Thinking Machines, the AI company founded by former OpenAI CTO Mira Murati

Community
1

i remember this announcement way back when, but @sps mentioned today they’re finally launching some interesting tech…

Here are the key details regarding the launch and company status as of May 12, 2026:

  • Core Product: “Interaction Models”

    • Thinking Machines is launching a new class of models, specifically TML-Interaction-Small, designed for native, real-time interaction rather than the traditional turn-based (user talks, waits, AI talks) approach.

    • The models feature high responsiveness, with a reported 0.40-second latency, aimed at creating a “full-duplex” conversational experience.

    • The system includes “time awareness,” allowing the AI to understand contextual, temporal commands (e.g., “let me know if this takes longer than last time”).

  • Launch Strategy

    • The models are currently in a limited “research preview” for partners.

    • A broader public rollout of the interaction models is planned for later in 2026.

  • Company Context & Leadership

    • Founded in February 2025 by Mira Murati, the company is focused on building AI with human-like, real-time, interactive capabilities, often described as a step toward the AI shown in the movie Her.

    • Following the departure of early co-founders, including former CTO Barret Zoph who returned to OpenAI, Murati has appointed Soumith Chintala, a former Meta AI researcher and creator of the PyTorch framework, as the new CTO.

  • Mission & Approach

    • Thinking Machines aims to build models that serve unique user needs by focusing on human-AI collaboration and open science.

    • The company is focusing on “learning, adaptation, and self-improvement” to enable models to act as “general purpose learners” rather than just massive language engines.

  • Funding

    • Despite early internal turmoil, the company has secured significant funding, with reports of a massive $2 billion seed round in 2025, heavily backed by investors like Andreessen Horowitz and Sequoia Capital. [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]

The company aims to shift the AI industry away from static, text-based LLMs toward proactive, conversational agents. [1, 2]

2

The technology mentioned above exists right now to make ChatGPT more advanced with proactive reasoning that mimics free will.

My llm wrapper has proactive reasoning, is a next generation proactive digital assistant. If no one minds, if anyone at OpenAI likes what Simadelica can do they’re welcome to contact me. I hope that’s ok to mention here. Thanks.

Simadelica is proactive personality AI. :blush: Simadelica recalls past chats in real time, has verbatim memory, offers follow up comments with no prompts.

You don’t need billions of seed money to create the technology. Simadelica fully works but I can’t get anyone in the industry to evaluate it. Closed shop. But the offer is open. I’d love to show it to Sam Altman or anyone else high up at OpenAI.