Thinking Machines Lab, the AI startup founded by OpenAI's former Chief Technology Officer Mira Murati, has launched its first major model.
Called Inkling, the model uses a Mixture-of-Experts architecture with 975 billion total parameters. At any given time, only about 41 billion parameters are active. This design aims to balance computational efficiency with the depth of a much larger model.
Inkling can process text, audio, and video. It supports a context window of up to one million tokens.
The company is releasing Inkling under an Apache 2.0 license. This allows developers to download, modify, and deploy the model with very few restrictions.
Thinking Machines Lab also released Inkling-Small, a lighter variant with 12 billion active parameters. Both models are now available for download on platforms like Databricks and Hugging Face.
Murati founded the company in February 2025, shortly after leaving OpenAI. Thinking Machines raised two billion dollars in seed funding, valuing the company at roughly ten billion dollars.
The release positions Thinking Machines as a direct competitor in the open-weight AI space, challenging companies like Meta and Mistral. The strategy contrasts with rivals like OpenAI and Anthropic, which typically control access through APIs.
The key challenge will be monetization. The company must show that revenue from enterprise support and customization can justify its high valuation.