thoughts on sonnet 5.5 after a day of real use

sonnet and opus 5.5 are clearly taught by the same teacher model (which i can only guess to be mythos 5.5 or even 6). their behaviors and personality are highly similar this time

but sonnet is definitely NOT “same as opus just 50% cheaper” - that’s true only for problems that don’t need much wisdom

there’s a clear difference when they are asked to propose plans for ambiguous product problems. you should be able to repro this by asking them to plan a complex enough feature whose direction is unclear

opus is able to approach problems with more strategic thinking “what’s the real goal here” while sonnet is more just looking at tactically “how do i get this done”

i also tested heavily using sonnet 5.5 as my firstmate and it’s working super well! i even prefer it over opus, because it’s faster, cheaper, and smart enough to understand my intent and handle the orchestration

my overall approach now is -
- opus plans
- sonnet implements
- once haiku 5.5 is here, i’ll use haiku for “fixes”
- if anything’s not going well, fable advises

sometimes i’ll go straight to fable for plans that i know requires its wisdom but this is becoming more and more rare in my experience (perhaps fable 5.5 will change it!)

i must say, anthropic’s latest line-up is incredibly compelling both in terms of quality and efficiency. we’ll see what openai cooked soon but they definitely have their work cut out for them