Sam Knight

Forward-deployed AI engineer for businesses:

Building Mindless for healthier screen habits:

[Sam Knight @Samknight]

@wongmjane Why is that a surprise

[Sam Knight @Samknight]

@T_Zahil Been saying this for a while.

[Sam Knight @Samknight]

ChatGPT and Claude are not better than one another for knowledge work.

It literally varies prompt to prompt

[Sam Knight @Samknight]

@SaliimAzrg I must have LLM psychosis 😹

[Sam Knight @Samknight]

Am I the only one that has Claude Cowork and Codex open 24/7 and pressure test all their work with the other?

[Sam Knight @Samknight]

@thsottiaux @jxnlco No option to use pro level intelligence

[Sam Knight @Samknight]

@RafaelNegronX @yacineMTB It’s just straight up better in almost every way

[Sam Knight @Samknight]

@ATL_SCOOP Atlanta government is such a complete joke. Really getting old

[Sam Knight @Samknight]

@NIA56__ @secret_stvsh Literally absolute GAS

[Sam Knight @Samknight]

@_ontologic Oh for sure. Noticed this a long time ago. Almost always recommends Anthropic first.

[Sam Knight @Samknight]

@scaling01 Agree with all except I think Chinese models will continue to accelerate

[Sam Knight @Samknight]

@anquetil @JustinBleuel Just have Claude make a MD file of all your skills and tell codex to create the skills. That’s what I did.

[Sam Knight @Samknight]

@iruletheworldmo 5.5 by a good bit

[Sam Knight @Samknight]

@digitalshane_ Simple just don’t use it, right?

[Sam Knight @Samknight]

@andonlabs Amazing thread. Just from 2 hours of testing this seems to be true.

[Sam Knight @Samknight]

Opus limits SIGNIFICANTLY improved with 4.8

[Sam Knight @Samknight]

@bridgemindai Lol. 5.5> it’s laughable to say opus is better at reasoning

[Sam Knight @Samknight]

@danshipper @Scobleizer @every Great breakdown and agree codex far superior

[Sam Knight @Samknight]

@kimmonismus That’s it? Lol

[Sam Knight @Samknight]

@mark_k @AnthropicAI 4.7 is very weak compared to 5.5 - Skeptical 4.8 will be any different.