News for hackers. Stuff that matters.
Re:Gemini: Google's most capable AI model yet by ZoomerCretin on Wednesday December 06, 2023 @07:32PM (#38551377)
From a quick test, it is not as good as GPT4-turbo at this leetcode problem: https://leetcode.com/problems/calculate-money-in-leetcode-ba...

Of the three answers Bard (Gemini Pro) gave, none worked, and the last two did not compile.

GPT4-turbo gave the correct answer the first time.

I agree that it is overstated. Gemini Ultra is supposed to be better than GPT4, and Pro is supposed to be Google's equivalent of GPT4-turbo, but it clearly isn't.

Re:Gemini: Google's most capable AI model yet by Davidzheng on Wednesday December 06, 2023 @08:05PM (#38551608)
GPT-3.5 is similar to Pro