@lord.luke5: There’s always a bigger fish in the sea. Every time we think we’re looking at the most powerful AI model in the world, something happens that reminds us we’re probably only seeing a fraction of what these labs are capable of internally. Astra looks insane. Then OpenAI is already talking about a next-generation model significantly beyond it. That’s the part people are missing. The public models might not be the frontier anymore they might just be the version of the frontier we’re allowed to see. And if the gap between released models and internal systems keeps growing, the next few years are going to get ridiculous. [OpenAI, GPT-6 Astra, AI models, frontier AI, internal AI models, AI agents, agent swarms, Navier-Stokes, artificial intelligence, next generation AI, AI research, AGI]
Luke
Region: US
Wednesday 09 September 2026 01:41:47 GMT
Music
Download
Comments
Eabusham2 :
One of my friends father works in ai training and he’s getting 500k+ offers it’s crazy
2026-09-11 01:41:21
1
Aiml3ssDE :
3.8 Flash is not agi wth
2026-09-09 05:39:20
25
ApplyGrid :
its OpenAi......
2026-09-10 17:20:53
1
𝕏0L0™ :
I honestly think Google is afraid to release it because they simply haven't done anything in the AI Race lately.
2026-09-09 20:34:35
5
joeywildeezsmokenuts :
The government has them especially the better GPT models
2026-09-09 02:56:48
9
ahdam,samad :
Google shouldn’t even be on the list, they release 3.9 mini flash next 💀
2026-09-09 06:12:29
6
Jay :
Just imagine the private stuff they got. we're getting the pub version. now take another step. imagine what the gov agencies have...
2026-09-09 03:04:57
23
user45 :
astra is the 2023 model from what the gov have
2026-09-09 04:25:03
4
omer :
3.8 True agi
2026-09-09 11:15:15
4
rusfine. :
Who is it?
2026-09-09 01:45:19
1
🅉a͟c͟h͟a͟r͟y͟ :
I remember 2/3 models. No effort selector, nowadays it’s like a launch control panel just choosing how many models and other settings 😭✌
2026-09-09 07:30:08
0
AI Fan Token — $AI :
Faces perform
2026-09-09 16:10:30
0
To see more videos from user @lord.luke5, please go to the Tikwm
homepage.