I find it confusing too. There are a bunch of different things that influence pricing (what's "auto" vs. "fast"?) and also "results"? Does that mean if I ask for 25 results, I pay $5 per 1000 requests, and if I ask for 26 results, I pay $25 per 1000 requests? or is that based on how many results Exa returns (which depends on how much data Exa has)? I'm sure if I understood the product (as your customers do) it would make sense, but as an outsider, it's not clear to me how much I would pay for the example searches (e.g: "space companies based in the US") or the demos.
Also minor notes:
* Research link in the footer is a 404
* Changelog link in the footer is a link to the blog
edit: after reading the documentation, there are 4 search types, "auto" and "fast" are 2 options that have been combined into a single column on the pricing table. The "latency" at the bottom is a description of the "fast" and "auto" options. I think I understand it now.
terraform sns to aws firehose example -- this one is not fair kinda -- only recently were you able to do this still the first result was a blog that had exactly what I needed
and
python decorator -- this one is timeless but I wanted to see new stuff
how am I to understand the search, context, answer, and research pricing? do I just look at search pricing if that's all I plan to do, for now?