I agree with you, but it's still a massive point though if they managed to go one step beyong in reasoning while keeping their efficiency, to take the most obvious one the token guzzling behavior of Fable is so big it makes as much headline as its heavy filtering and its actual great capacities when allowed to work.

If OpenAI is saying "oh by the way, upgrading to it won't make your usage massively explode", then I take that as a massive win. I'm on on Claude Max x20 and have accepted that Fable is not for me at its price point and token usage.