Thinking Machines Lab releases its first open AI model, Inkling

In short: Thinking Machines Lab has released Inkling, its first AI model that outside developers can download and change.

What happened

Thinking Machines Lab, a startup founded by former OpenAI CTO Mira Murati, announced a new AI model called Inkling. The company says this is its first major public release after about a year and a half spent building its systems mostly out of view.

Inkling is “open-weight,” which means the core files that make the model work can be downloaded by other companies and developers. In simple terms, it is more like getting the recipe and ingredients, not just ordering a finished meal. This is different from many popular AI products, where you can use the tool but cannot see or change what is inside.

The company says Inkling is built to be practical for organizations that want to adapt AI to their own needs. It can also warn when it is unsure instead of guessing, and it lets users adjust how much “thinking effort” it uses, trading speed for deeper work. Thinking Machines also says Inkling is not the strongest model available today, and it is aiming for balanced performance instead.

Thinking Machines plans to position Inkling as a starting point that customers can customize using its own platform called Tinker. The company also said Inkling was trained from scratch, but it used other open models, including Moonshot AI’s Kimi K2.5, to help create some early training material.

Why it matters

For regular people, this could shape how businesses use AI in everyday services, like customer support, coding tools, or internal paperwork. If more companies run AI they can adjust and keep in house, they may rely less on one shared chatbot for everything, and they may keep more of their business information private.

Source: TechCrunch AI