Fred Rivett because its will like cost effective more to use it for like larger document with little lost of intelligence

Makes sense, Michael! Will mention with the team

Hey Fred Rivett since gemini 2.5 pro is always down, can we just get 2.5 flash like its the second best thing in gemini model and like its good in term of price vs performance, and in leaderboard creative writing (what i care since i use lex for creative stuff) gemini 2.5 flash seems to be above claude 4 sonnet, i mean i know this leaderboard wont reflect real life usage and its probably wont be as good as sonnet but the prices is what i care more input just 0.30$/1m token and output 2.5$/1m compare to sonnet 3$ input and 15$ output since i use a lot of ai in lex if its cheaper i can use the ai more often

Hey Michael, thanks for the message! Unfortunately the main issue we have with Gemini is a low ceiling on our allowed usage from Google, which we keep bumping into. Adding more Gemini models wouldn't get around this issue.

Fred Rivett I see. That's sad, since there aren't many cheap models that are as good as Gemini Flash 2.5.

However, another reason I'm looking for a cheaper model is i like to use claude sonnet mainly version 3.7 and 4 but its so expensive

And I don't think any of the Claude models in lex are using caching. I noticed this by looking at the usage page noting how much credit i have and then comparing the cost of my first message with subsequent messages on my usage page:

If caching were on, you would expect a higher cost for the first message and a lower cost for messages that follow.

But from what I can see, that isn't happening with the Claude Sonnet models, and it's draining my tokens much faster.

Could you please enable caching for the Claude models? It would also be great if you could add a button next to the model selector to turn caching on or off so i can turn it off when not needed to save token usage.