Opus 4.7 vs. Sonnet 4.6

3 points by vdelpuerto 42 days ago | 5 comments

By standard I use Opus 4.7 for every task no matter what it is with Claude. But, because I was reaching almost 90% of the model use and with two days left before refresh I set Sonnet 4.6 for a simple task with my Price analyst agent. And it was kind of painful. 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 and it was not the same as with Opus. So, bottom line: not very sure if I can trust lower models with daily tasks where you need to delegate and stay confident it won't require too much iteration on those kinds of requests.

samuelknight 42 days ago |

The smarter the models, the less you have to structure the workflow. If you have a workflow that works well with a model today, you are in a good position because in a few months time you can switch to a much cheaper model.

One strategy I never had luck with was to have the smart model do an eval loop where it instructs the smaller model until it does the task. The problem is that the large model is way to prescriptive. This will improve in the near future because model labs are very focused on subagents

truetotosse 42 days ago |

So you are saying bigger newer better model is better than smaller old one in doing tasks?

vdelpuerto 42 days ago | |

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 42 days ago | | |

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 42 days ago | | |

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.