Thinking Machines Lab Releases First AI Model, Inkling
AI startup Thinking Machines Lab, founded by former OpenAI employees, has launched its first open-weight AI model, Inkling. The model can process audio and video alongside text.
New York – AI startup Thinking Machines Lab, founded by former OpenAI employees, has released its first artificial intelligence model, named Inkling. The model is open-weight, meaning researchers and other startups can download and modify it for their specific needs.
Inkling is capable of processing audio and video in addition to text. While the model may not top benchmarks in every specialized area, it performs well across many tasks and exhibits capabilities in advanced reasoning and coding. The model is large, with 975 billion parameters, and requires specialized computer chips to operate.
The release of open-weight models facilitates a more decentralized approach to AI development. The founders of Thinking Machines have expressed a vision where AI technology is not concentrated in the hands of a few companies. Open models allow more developers to build their own solutions.
Thinking Machines Lab was founded in February 2025 by former executives and researchers from OpenAI, including Mira Murati, John Schulman, and Lilian Weng. The startup secured the largest seed funding round in history, valued at $12 billion. The company has previously released a tool for model fine-tuning and showcased a feature for natural voice interaction.