Well if they are posting fraudulent benchmarks, that's a good sign to invest in their IPO. It's pure downside protection: IPO does well, profit. IPO does poorly, concrete evidence of pre-IPO fraud.

I personally don't think it's likely that OpenAI would post completely fake numbers in this pre-IPO period, but if you do, this is an opportunity.

I’m not an open-ai hater by default, I’m just waiting for someone to please explain how an incremental 5.5 cyber version is supposed to be already ahead of the flagship mythos model that’s been shaking up the software industry for a few months. If OpenAI had these supposed better-than-mythos 5.5 capabilities in their hands internally for some time now, why didn’t they make anything out of it in this era where everybody is desperate for any good press they can get?