Because only 5.5/4.7 level models are truely capable of "make no mistakes" type of work.
Local models need another iteration or two to reach that point.
I'd bet 2026 for sure tho.
@nicdunz Definitely has to do with fast mode something. Filled in lightning bolt is the 'look closely. Could be spark i guess, but I think 5.3 spark just said it rather than used an icon
I'm so fucking bullish on the United States and so dismayed that the average citizen is not so fucking proud that the best fucking AI companies are US companies
Let's get it together folks 🇺🇸
Sry if you came for the gentle homeschool poasts and got this instead 😅
“We’ve done the analysis, reusable rockets aren’t economic.”
SpaceX makes reusable rockets economic.
“We’ve done the analysis satellite internet isn’t economic. The antenna alone is tens of thousands of dollars. The cost to manage a constellation that size, the radiation, the space hardened solar cost…”
Satellite internet appears to be a very good business with antennas in the $100 range.
“We’ve done the analysis, orbital data centers aren’t economic. The radiators, launch costs, the radiation, the solar…”
You are here.
For the first time ever, bots and AI agents have surpassed human activity on the internet.
Our children will never know a world where Humans dominated the internet.
Welp, that happened faster than I predicted. Thought it would be end of 2027, then early 2027, but agentic traffic growing so fast that bots have now passed human traffic online for the first time in the Internet's history. https://t.co/2zX5bHdhsa
@aiseomastery@aisearchio No its not, and I for it! The two labs hellbent on spatial Intelligence and world models working together is how we start leaping forward.
You can increase the context in the config. But honestly I haven't run into issues with compaction since 5.5 released. It retains everything important.
Lots of suggestions here for "oldschool" work arounds like saving a running .MD file or opening new chats for new topics. I havent had to use those techniques in 5.5 at all with its new compaction logic. I just let it roll into megathreads and it works flawlessly.
I'm curious if youre using 5.5 xhigh when this happens. I often tell it not to make changes and it persists through compaction just fine.
@scottstts@scaling01 Aside from potentially mythos, there arent likely any models even 10T parms in size. 5.5 could be close but is likely more in the 7T range