[vdelpuerto]

Not really, although I appreciate the sarcasm. What I'm saying is that Anthropic and other providers tells you that you can rely on lower models for certain tasks that are usually daily tasks that don't require much more than basic understanding. To prepare a quotation doesn't seem like something that requires Opus 4.7 at max effort, don't you think? So the question is: what counts as a task you can delegate to a lower model and still trust the output?

[truetotosse]

It just had to look at an email from a client requesting a quotation, look for the database of our usual values, scope analysis according to the request

I don't think it is simple task for llm. Model needs to be tuned to go to places to fetch info, take actions.

[truetotosse]

Overall if you do not just try to oneshot the task and instead provide proper guidance where to look and what to do, you can achieve similar results with sonnet as well, you just need a better instruction.