@maria_rcks July 64th -> 64 + 7 =71 and “The number 71 represents a prime number in math, spiritual awakening in numerology”
Its all 4D chess you can’t see, ASCEND MARIA ASCEND!
@ThePrimeagen good for you, Gemini Flash 3.8 just launched lmao, didn't look too deep but seems to have good performance, good speed and mid to cheap price
One thing that I love about Anthropic model launches are these videos, short but really cool imo, they always find a way to make the video feel unique and not just a re-creation.
@scaling01 5 years is too long, 5 years is less time than ChatGpt age, architecture and techniques changed multiple times and reached heights we didn’t imagine.
Open sources will reach a local 240 layers within a year or 2 imo. It will find a way, not sure if it’ll matter much by then
Anyone remembers when Gemini 3.1 pro launched and it was so far ahead of competition and google were on top of everyone, less than a few weeks later everyone had their own "leap" and went above Gemini.
Today Gemini 3.1 is less interesting that 5.6 Luna. Any leap is exciting, the real exciting part for me is how the competition reacts, usually with a leap as well.
while it is impressive and its a bit sad for me that I had to leave them, the only downside I see is that its Fable and not Opus 5.1, meaning the token prices are double that of Opus and usually OAi releases after Anthropic with higher or same capabilities lately, I'd be suprised if Astra would be much further than Fable 5.1 and if it'll have a price increase from Sol.
@alexandr_wang Spark-1.2-contributor at 11T tokens means Meta got 11T tokens worth of data to improve their models, they better actually have a good leap using this data.