On July 15, 2026, Thinking Machines Lab announced the release of Inkling, its first model developed from the ground up. This innovative model is designed to handle complex multimodal tasks effectively.
Inkling boasts an impressive architecture featuring a Mixture-of-Experts design, comprising 975 billion total parameters, with 41 billion of those being actively utilized during processing.
One of the standout features of Inkling is its ability to manage a context window of up to 1 million tokens, allowing for extensive input data handling and improved performance in various applications.
The full model weights are made available under the Apache 2.0 license, promoting accessibility and collaboration within the AI research community.