I type a math question to research assistant
32%5
It gives results that have some markup and cites wolfram alpha. The markup should render to be legible.
Render the markup
Even if I pick expert on the Research, it switches me to Fast. And then even if I try expert again, it seems to show the cached result of the fast mode. So it's unclear if it's using expert. Also the Chat is defaulting to 3.5 turbo. I previously had Ultimate and this was not happening. I went from Ultimate to Starter to Ultimate and then it started happening.
Expect ultimate to default to expert or gpt4 models. Expect changing to expert model to not use the cached result of the fast model.
What happened? How did it happen? What are the steps to replicate the issue.
I tried pasting a long inpu into kagi assistant gpt4 chat and it failed with uri too long error. I pasted the same thing into chatgpt plus gpt4 and it accepted it. Can you please fix this to allow long inputs to the model
It fails because the entire text is part of the URL which has a limit I guess. It could be fixed by submitting in a post body instead.
What did you expect to happen? Describe the desired functionality.
It should have worked, not shown an error page