Basically, multiple submodels are part of GPT-5, and there is a model router that decides, when you ask a query, if the query requires one, or another, of the models to answer the query, for better performance. If you are able to disable the model router, but it is available or on by default, this market resolves YES.
If sufficiently ambiguous / unknown, may resolve N/A, but mostly if there is no given indication of this, the market will resolve NO.
Update 2025-07-24 (PST) (AI summary of creator comment): The router will be considered part of GPT-5 if it is available on either chatgpt.com or in the API.
An "auto router" will count if it routes between:
Different versions of GPT-5 (e.g., gpt-5-large and gpt-5-mini).
A version of GPT-5 and other non-GPT-5 named models.
@NielsW if the router is available either on chatgpt.com or in the api, i'll consider GPT-5 to involve a model router. an auto router that typically routes between different versions of gpt-5 would count, as would one that routes between versions of gpt-5 and other non-gpt-5 named models
@JaySocrates hmmmmm I am not sure, I will make a judgement based on how it is communicated at least in part. It would also depend on the specifics even if that description ends up matching what happens. I may N/A if it's sufficiently ambiguous but i don't think I"m able to answer whether that counts or doesn't in general