Will I judge GPT-5 to be smarter than o1 (not preview) after both are released?
6
Ṁ300
2026
63%
chance

Resolves subjectively, based on my analysis of benchmarks both official, third party, and my own.

Some examples of benchmarks I consider are MMLU, ZebraLogic, SWE-bench, simplebench, ARC, and livebench.

Some of my own evals are game-playing (tic-tac-toe, and connect 4), and creative writing (giving a model 3 random nouns and asking it to write a story involving them)

Get Ṁ1,000 play money
Sort by:

What do you do if no model named GPT 5 will be released, but instead they continue with the oN scheme for all their models?

@yetforever Resolves n/a. Though a departed researcher already described working on GPT-5 so I would be surprised if that happened