Exclusively: Mira Murate’s Stealth Ai launches its first product

Thinking machines laboratory, An young company that is highly funded by prominent researchers from OpenaiIts first product – a tool called Tinker to be automated to create dedicated AI models.

“We think (Tinker) will help enable researchers and developers to experience models, and will make border capabilities within the reach of all people,” says Mira Moratti, founder and CEO of thinking machines, in an interview with WIRED before the announcement.

Large companies and academic laboratories are already open source of Amnesty International to create new variables that have been improved for specific tasks, such as resolving mathematics problems, formulating legal agreements, or answering medical questions.

This work usually includes obtaining and managing groups of graphics processing units and using different software tools to ensure that the training processes are widely stable and effective. Tinker is to allow more companies, researchers, and even amateurs to adjust their artificial intelligence models by automating a lot of this work.

Basically, the team is betting that helping people to formulate border models will be the next big thing in artificial intelligence. There is a reason to believe that they may be right. Thinking machines laboratories are managed by researchers who played a fundamental role in creating Chatgpt. Compared to the similar tools in the market, Tinker is more powerful and easy to use, according to the Beta laboratory that you spoke with.

MURATI says Machines Lab Labs hope to remove mystery from the work involved in the world’s most powerful artificial intelligence models, and made it possible for more people to explore the external borders of Amnesty International. “We are making what is the possibility of a border within the reach of everyone, and this is completely changing in the game,” she says. “There are a lot of smart people there, and we need the largest possible number of smart people to do the AI ​​border research.”

Tinker currently allows users to adjust open models. Source: Meta’s Llama and Qwen’s QWEN. Users can write a few lines of software instructions to take advantage of the Tinker applications interface and start refining it through the learning subject to supervision, which means adjusting the form with the named data, or through Learning reinforcementIt is an increasingly common way to adjust models by giving them positive or negative reactions based on their outputs. Users can then download their dominant style well and run it wherever they want.

You see the artificial intelligence industry closely – partially due to the caliber of the team behind it.