They are not releasing O3 as a standalone model anymore. Not really sure what leap GPT 4.5 takes over 4o. GPT 5 will probably just be a UI change really. I think it's all marketing drivel now. They seem less excited overall about releasing any performance statistics about GPT 4.5. O3 they were enthusiastic about until DeepSeek R1 came out of China and blew away their cost and performance per watt.
Users of the free tier of ChatGPT will get unlimited chat access to GPT-5 at the standard intelligence setting; Plus subscribers will be able to run it at higher intelligence; and Pro subscribers will be able to run it at an “ever higher” level of intelligence, Altman said.
Sounds interesting that they will have a "Unified" approach where they want to simplify their UI wher you aren't choosing between GPT and O + (-mini) models series. Unlimited "standard intelligence" for free users I think, may just be the (-mini) instead.
I got to say I like O3, I like 4o. I like DeepSeek. I am actually for more excited and way more interested in smaller models that run locally and do reasoning. Those ones I feel are game changers now
I cheer on OpenAI, hope they consider being far more conscious about performance and open source. There will always been a need for 600 billion+ parameter models, most people don't even have the GPU compute to run 32 billion parameter models. There's no reason to be greedy and power hungry.
I hope Microsoft's Phi-5 will be the first of the smaller parameter reasoning ("chain of thought") models. I love Phi-4.