Hacker News
new
past
comments
ask
show
jobs
points
by
thefounder
45 minutes ago
|
comments
by
vunderba
38 minutes ago
|
[-]
I've actually settled on a very similar workflow - I mostly use Claude 4.6[1M] with adaptive reasoning disabled on High/Max for implementation, and then I'll do some combination of manual review in conjunction with GPT 5.5 xhigh.
reply