Overpromising will always lead to some sort of underdelivering, but what we're getting is still phenomenal.
If you got used to the complex OpenAI model naming and are confused with today’s GPT 5 release, here is the rough mapping:
GPT-4o → gpt-5 (main); o3 → gpt-5-thinking; o3-pro → gpt-5-pro.
gpt5-thinking as the default sounds right for me as well!
There was so much build-up by the team. I was expecting GPT 5 moving to a diffusion language model architecture or something. Instead we continue to exchange exponential compute for linear performance..
And wonder what happened to agent mode ?
it still exists! It just got better :) need to try it
If you got used to the complex OpenAI model naming and are confused with today’s GPT 5 release, here is the rough mapping:
GPT-4o → gpt-5 (main); o3 → gpt-5-thinking; o3-pro → gpt-5-pro.
gpt5-thinking as the default sounds right for me as well!
There was so much build-up by the team. I was expecting GPT 5 moving to a diffusion language model architecture or something. Instead we continue to exchange exponential compute for linear performance..
And wonder what happened to agent mode ?
it still exists! It just got better :) need to try it