{“toolCallComposer”:{“modelIntent”:“default”}}
Is what shows up in the billing section. Does anyone know how auto is being handled on the backend? Do i force it to use the model I want or is auto actually really good at selecting? Is auto always billed as a fast request or is it intelligent enough to when i do a relatively simple css change, say for rgb + transparency, to use a smaller model?
3 Likes
I hope someone from @cursor will respond, but because I’m paranoid about the quality of the agent, I always use sonnet 3.7 (non-max, non-thinking).
I hope i can answer your question as a fellow user.
You seem to have picked one of the threads where they havent answered because others contain more people that were interested and people kept reopening new threads for same topic..
Auto is what shows as default in your post..
Cursor team has clarified that is always a premium model (3.7 or similar others), that its a feature to allow users continue using AI when one or another provider is down or overloaded. etc…
To answer your question:
Yes its handled on their backend like all AI calls do.
You can either keep it on Auto or select your preferred model.
Auto would be a model thats likely what you want but since it cant know what you want it uses Auto.
No it doesnt use a totally simple model like 4o-mini or cursor_small. It uses premium models per Model list in Docu.