After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
I wrote about the state of AI, why I’m concerned about the next few years, and the choices we need to make to keep the future in humanity’s hands.
An Alien Mind: https://t.co/FeIfWNe0UE
I had early access to GPT-6. This model will completely shatter your understanding of what's possible, and it will start a new era of creativity.
It's hard to explain how big of a jump this is, so I'll share my tests.
1/6 Made this 3D model and animation in code from an image.
Codex usage at OpenAI gives us a preview of what agentic work may look like in the future.
In a new paper, the OpenAI Economic Research team looks at the broader shift from chat to delegation: people using agents not just to get answers, but to hand off longer, more complex work.
https://t.co/CQzNwpEevY
حوار معرفي يهم المشاهدين جميعا ليتعرفوا على كواليس الصحافة التليفزيونية يتألق فيه #جمال_ريّان ويقدم خلاصة خبرة 50 عاما من العمل في الإذاعة والتلفزيون أنصح الجميع بمشاهدته على المنصات الرقمية لشبكة الجزيرة الإعلامية https://t.co/fgMLEmSyZ5
توفي / جمال ريان
٧٤ عام
الدفن غدا الثلاثاء بعد صلاة العشاء في مقبرة مسيمير
عزاء الرجال يومي الأربعاء والخميس بعد صلاة التراويح في صالات ريجنسي بالدائري الرابع
https://t.co/vzqsHpPS0r
عزاء النساء في مجمع الفردان في الهلال– منزل رقم 130
https://t.co/5YuOxpxIzi
الله يرحمه ويغفر له
وفيات قطر
https://t.co/GdHFfAJMAs
GPT-5.4 can write Playwright code, read screenshots, and issue keyboard/mouse actions to operate computers.
You can steer its behavior and set custom confirmation policies for different risk tolerances.
On OSWorld-Verified, it achieves a state-of-the-art 75.0% success rate.
$SPY 690 target hit. I closed most puts and down to runners. I’ll probably close the rest around the 689.74 target. We can still overshoot towards the last target at 687.41 but I probably won’t wait for it.
My 6 most popular episodes this year are all practical AI tutorials:
1. Build an AI life co-pilot with Claude Code in 25 min with @AlexFinn: https://t.co/EgIc9Asvio
2. Master Cursor in 50 minutes with @leerob: https://t.co/B7wfEHecqM
3. Create beautiful designs with AI in 40 min with @mengto: https://t.co/6cwPz2sTrI
4. Build with multiple Claude Code AI agents with @kieranklaassen: https://t.co/FX496xvIy9
5. A 3 file system to vibe code production apps with @ryancarson: https://t.co/ABCMy5ySrI
6. Build a full stack app with Replit in 30 min with @mattyp: https://t.co/xWh2u1Kwcb
One thing is clear:
The guest's credentials (e.g., CPO, VP) don't really matter. What performs is practical workflows that you can apply right away.
📌 Subscribe to my YouTube for more interviews like the above: https://t.co/Ggqaa3F11Z
@OpenAI Y’all need to spend a day working on a decent CarPlay @ChatGPTapp experience.
We use you for some explanations and fact-checking while driving with the kid, but you still require fiddling on the phone/keyboard.
And we're live!
> codex -m gpt-5.2-codex
The vibes on GPT-5.2 are great, and we're excited to bring you the much-awaited Codex version. It's our best coding agent yet!
With significant jumps in our cyber evals, this will be our most careful, iterative deployment to date: