Hey there,

The biggest thing happening in AI right now is the fast advancement of open models.

Think DeepSeek, GLM, Kimi and the rest.

These are not as flashy as OpenAI’s GPT series and Anthropic’s Claude series. But they’re nearly as good and are about 1% of the cost, which is just crazy.

⬆️ Here’s what open models on American infrastructure cost. And they’re nearly as good as the frontier models! (Read all the details on our Private AI page if you’re interested for your own company.)

Now, I use Claude and ChatGPT every day, and I pay for them. BUT I also have major heartburn about data privacy, even for my relatively non-sensitive business data.

For folks who require data privacy, Claude and the GPT series are basically non-starters because the companies tell you directly that they’re going to log your data no matter what plan you’re on.

So if you’re in law, finance or any company where you need to keep your data out of third-party hands, OpenAI and Anthropic make no sense. (Anthropic just did a thing last week where they want to store your data for 5 years!