This is gold! Thanks, will take a closer look into this tomorrow. Not sure how I missed that. 

Tony

 [for instruction hierarchy evals]. We will follow up with a fix to improve these behaviors.

GPT-5 Is More Expensive Than Claude — Here’s Why

Hmm no. See https://stacker.news/items/1071526/r/optimism


> | Previous model | GPT-5 model         |
> | -------------- | ------------------- |
> | GPT-4o         | gpt-5-main          |
> | GPT-4o-mini    | gpt-5-main-mini     |
> | OpenAI o3      | gpt-5-thinking      |
> | OpenAI o4-mini | gpt-5-thinking-mini |
> | GPT-4.1-nano   | gpt-5-thinking-nano |
> | OpenAI o3 Pro  | gpt-5-thinking-pro  |

I think you'll want `-main`.

But note:

> We note regressions in performance for `gpt-5-main` [for instruction hierarchy evals]. We will follow up with a fix to improve these behaviors.

So you may want to wait this out a bit.

Previous model	GPT-5 model
GPT-4o	gpt-5-main
GPT-4o-mini	gpt-5-main-mini
OpenAI o3	gpt-5-thinking
OpenAI o4-mini	gpt-5-thinking-mini
GPT-4.1-nano	gpt-5-thinking-nano
OpenAI o3 Pro	gpt-5-thinking-pro