Unpopular opinion: LLMs are very shitty at writing code
I was at a party over the weekend and I talked to a couple guys both at top NYC quant funds. They told me they write all their code by hand and that even Fable’s code was terrible quality. They seemed to really dislike agentic coding.
This was really surprising to me. I am starting to believe there is a skill gap in using frontier LLMs that not everyone has.
It also made me realize how early AI adoption really is.
turns out we were comparing the wrong benchmarks, the ones highly skewed towards fastapi. we've been testing fastapi with 33 workers vs fastrapi with 1 worker.
in that matter, in the updated benchmarks, are coming soon!
frontend coming soon to FastRAPI !
great job @tiangolo and team, love this feature :)
With @FastAPI's new support for frontend you can now deploy full-stack apps to FastAPI Cloud 🏰
All in the same domain, no need to even worry about CORS 😎
fastapicloud.com/docs/builds…
