Positive users highlight Sol Ultra's value for complex multi-file coding and analysis tasks with large contexts, while negative users dismiss it as a placebo or agree the tests show no quality gains versus Xhigh.
Based on 13 visible X reactions from 53 accounts; directional sample.
Ask a question below.
Published answers will appear here.
I found one: i'm writing a super custom mac workspace manager and was having trouble getting around all the spaces and window manager limitations. Sol xhigh did not manage to find all the right private apis and nasty hacks necessary, but Sol Ultra did (it started reverse engineer a bunch of different applications and a bunch of other stuff I hadn't thought about). Surely, if i was more familiar with this corner of mac development, i probably could've guided a lower intelligence version to find the same. The hard part for me was the unknown unknowns.
@mitchellh I'm using it to adversarially review and improve sets of architectural design documents for software projects I didn't see myself tackling in the before time without at least 100M in Angel investment and a team of industry veterans, but the ceiling seems to be much higher. https://twitter.com/__eknight__/status/2075643450196971805
@mitchellh I've been using ultra for code-review of very sensitve code. Truthfully I haven't done a side-by-side but the number of edge-cases it has found is pretty astounding. Of course, many are unreachable in practice, but still.
@mitchellh im starting to think ultra is placebo for the confirmation bias crowd unless someone shows a side by side where it actually matters
@mitchellh i'm using ultra right now to do complex work that requires looking at a lot of files/infos together to gather as much info as possible
@mitchellh Skill issues
I haven’t successfully found a use case for Sol Ultra yet. I’ve run two days of side by side xhigh and ultra runs for planning and impl and I haven’t noticed any tangible quality change. I can see the difference in execution and token usage. Maybe I’m holding it wrong, but how?
@mitchellh No, that's pretty much right if you look at the pareto plot.
@mitchellh Did you try Max?
Positive users highlight Sol Ultra's value for complex multi-file coding and analysis tasks with large contexts, while negative users dismiss it as a placebo or agree the tests show no quality gains versus Xhigh.
Based on 13 visible X reactions from 53 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@mitchellh Skill issues