The Best AI Model In the World Is Free And Open Source?!

We are now a community of 234! Thank you❤️

This newsletter is free and I don’t use paid advertising. I completely rely on organic growth through users who like my content and share it.

So, if you like today’s edition, please take a moment to share this newsletter on social media or forward this email to someone you know.

If this email was forwarded to you, you can subscribe here.

If you want to create a newsletter with Beehiiv, you can sign up here.

OpenAI’s o1 model is one of the best AI models in the world. We don’t exactly know how it was trained and if copyrighted data has been used (wink wink) but it’s the best the AI companies have offered till date.

Imagine if we knew how it was trained - allowing other companies, teams, universities, to work better on their own models.

And, imagine if it was completely free and open source - enabling small companies and even individuals to self host the model and use it much more for a much lesser cost.

This was (kind of) the mission of OpenAI but as Elon Musk often says OpenAI has become more like “ClosedAI.”

Perhaps, the ‘imaginations’ mentioned above were bound to become a reality. No one expected it to be this fast but a Chinese company, DeepSeek, has made a model as good as OpenAI’s o1 and made it free and open-source!

DeepSeek R1

What makes the DeepSeek R1 even more remarkable is that it is trained through reinforcement learning, unlike other proprietary models which require a lot of human supervision.

Reinforcement learning is a method in which a model is rewarded for making good decisions and punished for making bad ones, without knowing which one is which. After a series of decisions, it learns to follow a path that was reinforced by those results.

Researchers from DeepSeek have shared that the model learned to reevaluate its initial approach to problems but it had never been programmed to do so. It developed the capability of its own through reinforcement.

The performance of R1 is impressive. It even surpasses the o1 model on some metrics.

But this is not all. The major win for DeepSeek R1 is the cost.

For a million tokens to be used by R1, you need $0.14.

The cost for the same in OpenAI o1? $7.50!

This makes the DeepSeek R1 98% cheaper than o1!

DeepSeek R1's code and training methods are completely open source under the MIT license, meaning anyone can use and modify the model without restrictions.

Simple conclusion? An unexpectedly fast leap in the right direction.

Reactions

NVIDIA’s top AI researcher, Dr. Jim Fan, noted how DeepSeek is doing what OpenAI was supposed to do.

There are also stories about how the Chinese media and government might be intentionally downplaying DeepSeek as a side project to portray how easy it was for the Chinese to achieve this remarkable feat. I can neither deny nor confirm this but we can enjoy the memes.

Also, DeepSeek is unique and other Chinese AI companies can be compared to their American counterparts - more hype than results.

Have you used DeepSeek’s models yet? What do you think?

There’s a reason 400,000 professionals read this daily.

Join The AI Report, trusted by 400,000+ professionals at Google, Microsoft, and OpenAI. Get daily insights, tools, and strategies to master practical AI skills that drive results.

Sign up now for free and work smarter, not harder.

Did you like today’s newsletter? Feel free to reply to this mail.

This newsletter is free but you can support me here.

I’d be happy to connect on Medium, LinkedIn and X.