For around a dollar a day, I am able to more than double my own productivity. What are the frontier models for/what are they trying to solve?
Are they designed to one shot applications? Is the expectation that we want to be able to "yolo" prompt LLMs and have them complete work unsupervised?
I have my IDE open and a prompt a given task with limited scope - like "write a function that does xyz" "add tests" "create an endpoint that accepts this json" "write a sql repository for this schema" etc.
I use Rust on the backend so if it compiles it works, meaning I don't need to worry about runtime errors - if it looks right, it's right. On the frontend I use Angular, which is about as constrained an environment as you can get in the FE world, so it takes a lot to lead an LLM astray.
So my requirement for high intelligence is pretty low and I am able to review diffs quickly. Deepseek flash is extremely fast so it can render my prompts quickly - while Haiku, Opus and Sonnet are slow and create the same outputs.
I'm guessing the value of Sonnet+ is less hand holding letting you give it larger scopes of work? Giving an AI a high level task and letting it do its thing without needing to review the output?
That might not be ok for your work / hobby / whatever, but if you wanna see the differences between models that's probably the best way.