Mira Murati's Thinking Machines Lab has released Inkling, an open-weight AI model with 975 billion parameters. The model is available under an open license, a move that could reshape the competitive dynamics of the AI industry by offering a powerful alternative to proprietary systems.
What Inkling offers
Inkling is a large-scale language model, but its defining feature is its open-weight license. That means developers and researchers can download, inspect, and modify the model's weights directly, rather than accessing it only through an API. With 975 billion parameters, it sits among the largest openly available models, though it's still smaller than some closed models from major labs. The open license allows for customization, fine-tuning, and integration into a wide range of applications without the restrictions typical of proprietary models.
The release challenges the prevailing trend toward closed, API-only models. Companies like OpenAI and Anthropic have increasingly restricted access to their most powerful models, citing safety and competitive concerns. By contrast, Thinking Machines Lab's approach could spur innovation by giving developers more freedom to experiment and adapt the model. It also lowers the barrier for smaller players who cannot afford expensive API calls or lack the data to train models from scratch. The move may pressure other labs to reconsider their openness strategies.
With Inkling now publicly available, the focus shifts to adoption. Developers can start building on top of it immediately, and the lab expects a wave of community-driven improvements and applications. The broader AI market will be watching closely to see whether open-weight models can compete effectively with closed systems in terms of performance, safety, and commercial viability. No further announcements from Thinking Machines Lab have been made yet, but the release itself sets a new benchmark for openness in the field.
