HLE value here is calculated simply as:

But if we are getting more value, then it's worth the increasing cost. Value here is measured by intelligence (HLE score) divided by the cost per model. The value for the flash models has been going down while value otherwise has gone up!

The Google Gemini models vary widely in price and intelligence. Here's a quick analysis to help you choose the right model at the right value for your use case and see the industry trends. One thing is for sure: the cost of AI via the API is going up dramatically over time!

1st step completed.

He's been right this whole time. The big labs have been far more risk averse and uncreative than anyone can imagine. Luckily, we're not a big lab at DeepWriter. A team of 3 with under $100k doesn't beat a team of with billions of dollars in R&D by doing what everyone else does.

Our HLE score proves it. No LLM can score 50.91. They need a lot of help!

Those in the robopsychology movement have known this for a long time. IYKYK. At DeepWriter, tapping into vast latent spaces is one of our trade secrets. Why only use 10% of the capabilities? And part of why we beat the labs with billions of dollars for R&D.

We used Gemini 3.0 Pro primarily with Gemini 2.5 flash for low level backend.

We're working on this.