Andy Halliday

We had an AI bubble pop, a small bubble pop early in the year that was partly a function of the emergence of the Chinese low-cost models being competitive at the frontier level.

But more importantly, it had to do with the questions about whether enterprise, who's going to ultimately

pay for a lot of this in long-term remuneration to the developers of AI, whether they were really seeing the results necessary.

And I think it was just a little too early to see that that would happen.

And I'm long-term an optimist about just how completely AI will transform the economics of business.

And we can see that on a micro level in many ways right now.

But here you have these Chinese companies.

And Moonshot is a really impressive development.

And Kimmy K2 thinking is something that if you are using AI in the API world, you're developing something, you ought to be taking advantage of that.

Why?

Because it's a fraction of the cost.

of the major frontier models like if you're if you can get access to these things like on open router or there's a cool new tool that i'm going to tell you about right now called llmgateway.io

llmgateway.io.

You can get to any model of importance across the spectrum, closed and open, and access them via a single API at this LLM gateway.

So this is simplifying dramatically the switchability, if you will, of models for comparison purposes or for cost reductions over time.

You build your application to access AI tools and models through this single API.

And you can access multiple models simultaneously in your process by scheduling the API calls however you like.

And if you notice in LLM Gateway that, oh, this model does just as well as that one, well, let's switch over to that one because it's a fraction of the cost.

Kimi K2 thinking is available there along with all the other different models from Moonshot.

OpenAI models are there, Gemini models are there, everything.