swarek🦍
6.9K posts
swarek🦍
@Swarek_
vibecoder - https://t.co/x0RzDQ0d4F
Marseille, France Katılım Eylül 2014
185 Takip Edilen2.9K Takipçiler
viens on parie 200 euros que jmet ca en ads et ca rapporte pas UNE SEULE vente
F@Farescrbs
L'IA peut faire ça stv
Français
Beyond today’s instant model update, what can we improve?
Sam Altman@sama
hey chat, we haven't forgotten about you 👀
English
@rileybrown I'll be on the next cohort for the testing if you want some prompts
English
I hardly ever say this...
but my spidey senses are tingling...
something is off about this.
mark my words.
Alexander Whedon@alex_whedon
Introducing SubQ - a major breakthrough in LLM intelligence. It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA), And the first frontier model with a 12 million token context window which is: - 52x faster than FlashAttention at 1MM tokens - Less than 5% the cost of Opus Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention). Only a small fraction actually matter. @subquadratic finds and focuses only on the ones that do. That's nearly 1,000x less compute and a new way for LLMs to scale.
English
5.5 instant comes to ChatGPT today!
imo it is a pretty big upgrade, i really like using it.
Eric@ericmitchellai
Excited that we're updating the default model in ChatGPT today! 5.5 instant is a substantial improvement in intelligence, image perception, and factuality. It also updates the writing style to be a bit plainer and more straightforward. What was on your wishlist?
English
Introducing SubQ - a major breakthrough in LLM intelligence.
It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA),
And the first frontier model with a 12 million token context window which is:
- 52x faster than FlashAttention at 1MM tokens
- Less than 5% the cost of Opus
Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention).
Only a small fraction actually matter.
@subquadratic finds and focuses only on the ones that do.
That's nearly 1,000x less compute and a new way for LLMs to scale.
English
@alex_whedon That's amazing ! Just requested access, so hype to try it. You sell it for code, but is it interesting as a personal assistant ? Like natural speaking
English
English
Juice values of GPT 5.5 in Codex
low - 12
medium - 24
high - 96
extra high - 768
same for standard and speed
Harshith@HarshithLucky3
Juice value of GPT 5.5 thinking standard is 24 GPT 5.5 thinking extended is 192
English
OPENAI 👀: Codex app got updated to version 26.429.20946 with loads of minor tweaks and new hidden additions!
What's hidden? (not available yet)
- A new Remote Control feature and Connections section in settings
- Codex "Avatars" got renamed into "Pets"
- A new Keyboard category in settings for shortcut management
- A new onboarding widget prompting to connect core integrations with email, calendar, and files
What's new?
- Updated onboarding flow
- 20% faster computer & browser use
- Upgraded slides and sheets capabilities
- Annotation for browser, artifacts, and code
- Cleaner design & performance improvements
Additional changes?
- It seems like Browser Use got disabled in EU
- It seems like Computer Use got disabled in EU
Sam Altman@sama
big upgrade for codex today! try it for non-coding computer work.
English
@flowersslop @sama What does it misses for openclaw ? Real question, I never use openclaw but with the heartbeat in codex it's so cool
English
big upgrade for codex today!
try it for non-coding computer work.
English
@Reshi_18 @KushinaGuccii Ma base morale : réduire la souffrance évitable et respecter le consentement, l’autonomie et l’intégrité des personnes. Le viol viole tout ça à la fois. Donc oui, il est mauvais sans avoir besoin d’un dieu pour le décréter
Français
@KushinaGuccii Les athées vous avez quel base morale objective pour affirmer que le viol est factuellement mauvais?
Français
Mais quels athées pensent ça ? J’en connais aucun comme ça ptdr
regarde moi qd tu me leche@legaroa2pm
La religions : soyez pur et respectez vous Les athées : le viol n’est pas un crime Stupide goyim mentalité 🤦🏽🤦🏻🤦🏻🤦🏿♂️🤦🏽♀️
Français
the results of the GPT 5.5 reasoning level thread:
1. medium
2. xhigh
3. high
4. low
i’m surprised there was a lot more xhigh than high
Anthony Kroeger@kr0der
now that it's been ~1 day of GPT 5.5, what reasoning level are you guys sticking with?
English
how easy is to be forgotten 🥀
i guess remodex had a good run
but i want to see how much it can expand
Dan McAteer@daniel_mac8
when codex gets an ios app the world will never be the same
English
Codex users: I'm curious how you feel about its personality/communication style. When do you want it more concise vs more explanatory? More direct vs more collaborative? Examples where it talked to you especially well, or badly, would be gold. DMs open!
English