@oprydai Had to fit an extra desk in my room
Just started learning robotics from absolute zero
3d printed an arm and starting to train it for real world use cases
+ working on massive distribution systems w phone farms
Only reason i keep paying claude is fable. just superior than any other model rn
Gpt5.6 ultra disappointed me tbh expected it to feel more intelligent it didnt
K3 is doing decent results not a white pass for everything but the flagging of what it wont do got really reduced
Opposite of fable, which just flags i have to keep doing prompt engineering and compacting / sending handoffs to fresh sessions to bypass the flags
Grok4.5 is just super fast not very intelligent yet
@SpaceXAI are the ones im most excited about
Elon is accelerating on his lane for the long term win
Finished the robotic arm
Hardware side closed
Now starting to train the ai to operate in the real world
Learning robotics from absolute zero
Sharing all the journey in public
I have the confidence that I can do absolutely anything nowadays.
Iβve proven it over and over to myself.
I get into challenging completely new projects where I have absolutely no knowledge.
And just figure everything out on the way. I just deeply trust in my capabilities and know I will solve/learn anything one way or the other.
copying apps that do +1M$/month.
first one just hit app store review
no original ideas needed.
take what already prints money and ship it
two more into review this week
we got massive distribution covered internally already
@minchoi Our 2T model, which is better than our 1.5T in every way, will finish initial training next week. It might be able to exceed Kimi, but with speed and token efficiency close to our 1.5T (aka Grok 4.5).
Testing k3 limits trying to reverse engineer a certified third party app to have it fully custom rebuilt
Letβs test how far the (non) restrictions get in this model