Google’s Gemini is Forcing OpenAI’s Hand with GPT-5 — Your Move, Apple!

Gemini Ultra’s performance isn’t a shocker but it’s enough to make you think twice about OpenAI — let me share a few thoughts after today’s announcement.

Let’s start by looking at some of Gemini’s benchmarks:

Ultra is the first model to achieve human-expert performance on the MMLU exam benchmark, and it also achieves state-of-the-art results in 30 of 32 benchmarks, which is truly impressive.

Join our next cohort: Full-stack GenAI SaaS Product in 4 weeks!

On the other hand, despite Google pouring all its resources into Gemini, the fact that it didn’t significantly outperform GPT-4 might indicate we’ve hit a plateau with Transformer models. This suggests we may need a new breakthrough in our approach to achieve a significant leap in performance and abilities.

I bet you have seen some of the demo videos, Gemini is absolutely nailing it across images, audio, video, and text. The teams at Google DeepMind and Google Research crafted a masterpiece.