Bessi
2.8K posts

Bessi
@LLMpsycho
I'm just a coder for fun, like Saitama, suffering from LLM psychosis.
Albania Entrou em Temmuz 2024
1K Seguindo996 Seguidores
Bessi retweetou

I've never felt such appreciation for another country - Japan is epic.
Sassafrass84@Sassafrass_84
I'm deceased. Japanese X is wild, y'all.
English

@HarryStuck77 @droid Latest updates have improved it further. I'm at that point where i have everything to code, i dont anything else.
English

A professional approach starts with a lot of planning. @droid is on top of this game. Not just a plan and a task list. A proper validation contract.

English
Bessi retweetou

openai’s new model spud isn’t just an incredibly powerful and chonky new model.
it’s that as well but
it’s a a refocusing of their attention on what really matters. serving people with an always on intelligence that can serve them.
they’ll pull together agent mode, atlas, deep research, codex, etc etc. into one unified agi experience with a model that feels alive and vastly more capable than its competitors
people at openai are shocked enough at the progress of this model they felt comfortable wiping out all the efforts made with sora and any billion dollar disney deals
you don’t create an ai deployment division. unless. you. built. a.g.i.
let that sink in gremlins.
English

@LLMpsycho I’m still having problems in long missions at least with 5.4 but there’s a work around
English

We only hire builders (and we’re on a hiring spree)!
Reply with something you've built. I'll read them personally. We’re interviewing the best ones.
You’ll be a good fit if you:
- work best without permission
- default to “how could I automate this”
- had weird teenage hobbies
- spend your sunday making side projects
- have more Claude agents than cousins
- shipped something this week
- make prototypes, not powerpoints
- don’t like hierarchy
- are good at games: chess, monopoly, poker
- would take dinner with Elon over $100k
Good luck,
Eric
English

@droid @FactoryAI Is going to win and its not even close.
I was trying EVERYTHING to fix my holdout RMSE on a model I am working on and it kept getting caught over and over and even regressing.
One prompt using droid + GPT5.4 High and fixed it. I spent almost $100 in Codex credits trying to fix this before hand.
If i could I would buy the Max plan TODAY

English

@HarryStuck77 Mine not finished yet, will share the total when it does.
English

@thsottiaux this is the part that makes codex feel useful in real work
English

Boom. Plugins! Codex now has access to most things it couldn’t reach before and all super easy behind the same with you use for your ChatGPT account.
I use this for *everything*, from managing my calendar, triaging incoming bugs, keeping up to date with what everyone is doing at the company and so much more
OpenAI Developers@OpenAIDevs
We're rolling out plugins in Codex. Codex now works seamlessly out of the box with the most important tools builders already use, like @SlackHQ, @Figma, @NotionHQ, @gmail, and more. developers.openai.com/codex/plugins
English

@petergostev tracking real repos makes this feel a lot more useful than pure hype.
English

Claude vs Codex - who is winning? The best data I could find to do a fair comparison between the two. It is not ideal, but some rough indication:
- Usage: NPM and Homebrew installs
- Footprint: repos with .claude/ or .codex/
- And of the repos how many are being actively updated
So pretty clear that Claude Code is leading - whether it is 70% of 95% of the this slice of the market is not clear. But at least both are being used actively in real repos.
If you have any other ideas for what sources to track let me know.

English








