Would be useful to know how much each summarizer usage cost me. Particularly since 3 of the models are per-token, so it varies depending on the size of the page, which isn't easy to quickly. As well, for long pages, it would be useful to compare the cost of the the per-token models versus the fixed-cost one (Muriel).
This could be implemented in several ways. In increasing order of usefulness:
- add a line item on the billing page for each invocation of the API (including costless ones) so I can look up the entire history of my usage. not super useful, since I probably won't visit there often, and I may forget what I looked up and why. showing the url and timestamp of the page would help with that.
- add feedback after clicking summarize. something like "your summary just cost $x.xx!" so I can see it right after I click it. more useful, since it would put the info in my face every time I use it. downside is, I may forget or click away, then want a way to look it up afterward.
- add a preview of the cost before clicking summarize. automatic would be nice. maybe triggered by pulling up the extension, or when switching models? alternatively, an "estimate cost" button would do the trick. showing the cost of different models at a glance would also be nice to have, so you can see "oh I should switch to Muriel for this summary, the article is longer than I realized!" without having to manually jump through hoops (which I probably would not do most of the time. people are lazy.). a fun option would be a checkbox like "use Muriel if cheaper".
Related, but you clearly cache these summaries. I noticed when I clicked a second time on the summarize button, and the result showed up right away. Do cache hits still cost me money? If not, telling me which models are cached or not could save me a lot of money. And then a "use cached response, if available" would be invaluable. (or maybe "use cached response, if available in these models:" with checkboxes for the models you want to enable.
