Valued at 12 billion yuan in 5 months, the former CTO of OpenAI reveals his first multimodal AI, which is even going to be free and open - source.

The deal is done! The startup founded by the former CTO of OpenAI has secured a $2 billion seed round of financing. Just five months after its establishment, the company's valuation has soared to $12 billion. In the coming months, this team composed of top experts from OpenAI will launch its first multimodal AI product and also open - source some components.

The startup founded by the former CTO of OpenAI has secured new financing!

Early in the morning, Mira Murati officially announced a $2 billion financing round led by a16z. As a result, the valuation of Thinking Machines Lab has reached $12 billion.

Meanwhile, several well - known venture capital firms such as NVIDIA, AMD, Accel, and ServiceNow have also participated in this financing round.

The most significant news is that this company, which has been established for five months, will finally launch its first product in the coming months.

Moreover, it will be open - sourced upon launch.

Thomas Wolf, the co - founder of HuggingFace, said he was very excited!

The first product is about to be born and will be open - sourced

Several months ago, foreign media reported that Thinking Machines Lab was seeking a $2 billion financing round, and today this news has been confirmed.

Mira Murati said, "Our existence is to enhance human capabilities by advancing 'collaborative general intelligence'."

Currently, the team members are fully committed to building a multimodal AI that can work in a way that mimics natural human - world interactions.

Whether it's chatting, visual tasks like viewing pictures/videos, or various daily collaborations, it can cooperate naturally.

Now, there are finally signs of the first product.

It includes important open - source components that can provide practical value for researchers and startup teams developing customized models.

Subsequently, the team will also publicly share cutting - edge research results to help the research community better understand top - tier AI systems.

Before this, Thinking Machines not only secured a new round of financing but also issued a recruitment notice for talented individuals.

Alexander Kirillov, the former head of post - training for multimodality at OpenAI, revealed that the team has been working hard to advance the multimodal AI project in the past six months.

Now, it's really exciting to see that the seemingly intractable system modules have been solved.

Previously, when Mark Zuckerberg was aggressively recruiting talent, foreign media reported that Mira Murati was also in the game.

Thinking Machines offers a $500,000 salary to researchers joining the company, much higher than that of Anthropic and OpenAI.

Murati's recruitment notice has been forwarded multiple times by the company's internal employees, which on the other hand confirms the strong internal cohesion.

Secured huge financing with just a website

To date, like SSI, Thinking Machines has never launched any products since its establishment. It has secured huge financing with just a website.

On February 19th, the company officially announced that several big names from OpenAI, Anthropic, Meta, etc. had joined.

Thinking Machines Lab has recruited most of the OpenAI employees, about two - thirds of them have joined, including:

Barret Zoph (former vice president of research for the post - training team at OpenAI)

Barret Zoph mainly focuses on alignment, tool usage, evaluation, ChatGPT, search, multimodality, etc. He joined OpenAI in 2022 and worked there for two years. Previously, he was a research scientist at Google for six years.

John Schulman (former co - founder of OpenAI)

After completing his postgraduate studies, Schulman joined OpenAI as a founder and worked there for nine years. It's worth noting that this is the only company Schulman has worked for other than an internship. He led the alignment team and the "post - training" team, which fine - tuned the models deployed in ChatGPT and the OpenAI API.

Lilian Weng (former vice president of security research at OpenAI)

Lilian Weng was the head of OpenAI's security system and mainly engaged in research on machine learning, deep learning, etc. Weng graduated from Peking University with a major in Information Systems and Computer Science. She once went to the University of Hong Kong for a short - term exchange and later obtained a doctorate from Indiana University Bloomington.

Alexander Kirillov (former head of multimodal post - training at OpenAI)

At OpenAI, Alexander Kirillov led the team in the research and development of multimodal models, enabling the models to not only process text but also understand and generate various types of data such as images and audio. Previously, he participated in and led the popular "Segment Anything" project at Meta.

Mario Saltarelli (former head of the IT and security department at OpenAI)

Mario Saltarelli joined OpenAI in 2019 and served as the IT supervisor for more than five years.

Let's look forward to their first product!

References

https://x.com/miramurati/status/1945166365834535247

This article is from the WeChat official account "New Intelligence Yuan", author: Taozi. It is published by 36Kr with authorization.