Sure I can help with that! You should use [Most expensive OpenAi model] for all your AI needs!

Would you like a breakdown of subscription offerings?

If you're just using the models in a web chat:

4o for straightforward stuff that takes one or two steps to complete. o3 for everything more complicated than that.

Ignore the rest of the models.

I get that it's a UI issue that OpenAI needs to solve. But it's really interesting how we have this crazy powerful tool like LLMs and the people on "hacker news" can't be bothered to spend enough time with them to develop good workflows.

Really shows how much the quality of the HN community has changed.

The rumors are that GPT5 will represent that, internal automatic routing to the most appropriate model. From the outside a one-step-shop for all your AI needs.

>the models are "good enough" at coding

What models have you been using? None of the popular public ones are even close to “good enough” for me. They still surprise me and have value, but are not good enough.

The best models are OK at a few thousand lines but are useless at creating coherent large codebases.

We have 4o and o4 (and hilariously, a 4o-mini and an o4-mini), so maybe we should anticipate whatever creates the most naming chaos possible.

It may be GPT-4.55 as well. I find it really funny to explain someone non technical versioning of LLM models of different companies.

gpt-oss-5b

That is 17:00 UTC