@nextpath.ai: its next-generation model and what the company is calling a "generational leap" in Al. The biggest shift is that Astra is built to do far more than answer questions. It can work across software, browse, code, use engineering tools, handle long-running tasks and coordinate agents on complex work. OpenAl also says the jump from GPT-5.6 Sol to Astra is bigger than the jump to Sol from its previous models. The full benchmarks numbers are absolutely insane. Astra reportedly scored 98.6% on ARC-AGI-3, 97.6% on FrontierMath Tier 4 v2 and 100% on ExploitBench, while Greg Brockman said he personally believes OpenAl has reached AGI and ended the briefing with: "Welcome to the AGI era." Rollout starts with approved users in OpenAl's Daybreak program, with paid ChatGPT users and API developers getting access over the coming What are your thoughts on this?

travelwithmufasa :

What is AGI

AnhedonicLobster :

Tired of the fake hype over AI. it's mid

Justin Griffith :

They keep moving the AGI bar but we’re already there. The current models are insane. Astra is going to be a game changer.

Ninno :

Right before IPO yeah sure buddy 😂 cash out and run fellas this ain’t movin shit

Dio :

no ai please

3y3less :

AGI... yeah sure... Another slop

Isaac :

Astra, write code for GPT-7 and make it 10x better, make no mistakes

ryonic82 :

This is what you call cheating on the exam. Why are all the rest of the benchmarks largely in line with Claude?

I’m your father :

Yeah tomorrow I am releasing something that beats all benchmarks just wait

BitcoinReaper :

Uoopy Nan 🎀 :

AI will never be as a smart as a humans especially me

Danny :

it's far from AGI 😂

Saturo :

They did say nothing abt Agi🥀 Ai as we have if rn can never become an Agi no matter how much you train it, the code will literally not make it possible

dml10181 :

Oh they just happened to create AGI months before IPO launch

The Great Grahambini :

98.6% on ARC-AGI3 is insane if there are no caveats to this score, but I suspect there are caveats

Colin R Mah 🇨🇦🇺🇦 :

I swear the image winked at me lol

sonboku56 :

Agi achieved. Again. 😂😂

Jotinder Jot :

Gpt 6 before gta 6

ScottishPerson🏴󠁧󠁢󠁳󠁣󠁴󠁿 :

You won’t get AGI out of LLMs.

jay o :

cheated on every benchmark 🥀

nvxqdgjkoui :

this is not AGI

The_Friendly_Guy :

it's not agi yet, but it's impressive. close to fable 5 before they lobotomized it

Ryan E. Brown :

How does it compare to Deep Seek?

grozen :

GTA 6 is insane!

ajouter un nom :

the usual ✨pls pump my bag✨ by altman