I think it at least is somewhat analogous to what happened with pricing on previous models. GPT 4, despite being less capable than 4o, is an order of magnitude more expensive, and comparably expensive to o1. It seems like once the model is out, the price is the price, and the performance gains emerge but they emerge attached to new minified variations of previous models.
I think it at least is somewhat analogous to what happened with pricing on previous models. GPT 4, despite being less capable than 4o, is an order of magnitude more expensive, and comparably expensive to o1. It seems like once the model is out, the price is the price, and the performance gains emerge but they emerge attached to new minified variations of previous models.