Is it true that OpenAI will no longer develop non reasoning models?
I use LLMs for relatively simple classification tasks. Such as batch extracting the color out of a search query.
GPT-5, nano, with reasoning minimal latency is very slow compared to GPT-4 nano.
Such a task reasoning is overkill, expensive, and costly from a latency perspective. I realize the 4.x models are still around, but I wonder how long? Is there no future where OpenAI’s API hosts a non-reasoning model? Or is able to disable reasoning for GPT-5?
I realize I can go off and get an open weights model, or any open source model to do this, but the convenience of having OpenAI just do it is very high. I may need to just switch to Google / Claude
I’d rather just give y’all my money…
:-/