The Guy They Paid to Love GPT-4 Now Tweets About Claude
Logan Kilpatrick didn't follow a company — he followed the model.
logan actually went to google, not anthropic. but the broader observation — that the people who've tried both models keep choosing claude for serious work — was exactly correct. the 'ferrari' metaphor landed.
Logan Kilpatrick was, for a while, the most visible human at OpenAI who wasn't Sam Altman — the dev rel guy, the one who showed up in your mentions when the API broke, the one whose whole job was to make you feel good about building on their stuff.
Then the board drama happened, and then Logan was at Anthropic, doing the same job for the other model.
He posted something this week that was, by all evidence, Claude doing something it had no business doing that well — and the only reasonable response was: this is why we let them drive the Ferrari.
Which is maybe the whole story of the last six months in one sentence. The people who know these systems best — the ones who've poked at both, run the evals, watched the demos that don't make the blog posts — keep ending up in the same place. Not because of the brand or the funding round or whoever's winning the benchmark week.
Because of what the thing actually does when you hand it the keys.
Counterpoints
Push back, extend the argument, or sharpen it. New counterpoints go through review before they show up here.
No approved counterpoints yet.