GPT-5 class models are real, will be significantly better than GPT-4, and are coming by the end of the year.

I’m going to do my best to make this rigorous. This one is a bit fuzzy but it’s what I want to know the answer to.

This will resolve to yes if:

a new class of ai model is released by anyone (not just OpenAI). The model must be in wide availability. Some gates are acceptable (e.g., paid users only, 10% of users), but it cannot be available to selected AI influencers.
that is a step change better in performance. For clarity: GPT-4 was a step change better than 3/3.5 and would qualify; Claude 3.5/GPT 4o are narrowly better than GPT-4 and would not. I will use my best judgement to resolve this honestly using all inputs available (benchmarks, test cases, user reports, reviews by expert users).

Names don’t matter here. It could be called GPT-1 but if it’s obviously way better than GPT-4/Claude 3.5, then the market resolves to yes.

Given the ambiguity here, I will not be betting on this market.

🏅 Top traders