Chris_Online

AI fanatic | AI efficiency Hacks | EV + AI Adventures | Cats, Dogs & Tech etc..

[Chris_Online @Mista_AI]

Grandma bought 60 shares of $SPCX at $165. Just bought another 40 at $132 to average down. Now holding 100 shares at $148 average. Anyone else averaging down on this dip? $SPCX #SpaceX #Starlink

[Chris_Online @Mista_AI]

Starship Flight 13 is happening today with the first real Starlink V3 satellites on board. A successful flight proves the upgraded vehicle works and accelerates Starlink growth. That’s direct revenue and a huge catalyst for the stock. $SPCX
#Starship #SpaceX #Starlink #SPCX

[Chris_Online @Mista_AI]

@elonmusk Quick idea on the EchoStar/Dish bankruptcy assets. Their ~24k tower sites with radios are going to auction in mid-August for a stalking horse bid of just $300M. You already committed nearly $20B for that exact spectrum. Buying those assets privately before the Aug 10 bid deadline would be an absolute no-brainer — cheap infrastructure already in place that perfectly matches your spectrum. Even if you don't run a full ground network, you could hold it as an asset and sell it later. Smart move before the big stock unlock after earnings. What do you think?"
Best hashtags to add at the end: #Starlink #SpaceX #Starship #EchoStar

[Chris_Online @Mista_AI]

I asked Grok to tell me how it compares to other versions and why it is better. See my conversation

[Chris_Online @Mista_AI]

Grok 4.5 lis Live. Check it out.

[Chris_Online @Mista_AI]

I just tried to use https://t.co/gIbopbtSVp to get the limites and this was the result. Not very descriptive

[Chris_Online @Mista_AI]

@testerlabor This has been hyped for far too long. We had a saying in the Military : "Don't tell me about the pains, Just show me the Baby. " I'm all in on Gork and all things Elon, buy it's only in retraining, that means it won't be ready for 6 or so months.

[Chris_Online @Mista_AI]

WOW! check this out, anyone in my shoes ? @Grok what do you think about this ? @elonmusk @xai can you shed any light on this. I know this is Not Breaking News

[Chris_Online @Mista_AI]

Wow! Supergrok & Voice are hitting limits quickly, In hindsight, the move looks riskier than it did at the time. They got money and focus for Colossus 2, but they clearly underestimated how fast voice and SuperGrok demand would grow. If keeping Colossus 1 running for inference would've prevented these tight limits, it probably would've been smarter to hold onto it for a while longer.
Right now it feels like they traded short-term user experience for long-term infrastructure — and the short-term pain is hitting paying customers hard. That's tough to justify.

[Chris_Online @Mista_AI]

@Figure_robot This poor intern is going to have to go back to school later and when the teacher asks him what did you accomplish over your internship , he can reply "I MADE HISTORY"

[Chris_Online @Mista_AI]

@Figure_robot Looks like the human is trying to give more attention to the flattering of the packages ( perhaps trying to go slower ) mmmm, but it looks great ( GREAT MARKETING )

[Chris_Online @Mista_AI]

@Figure_robot Nothing more satisfying than watching someone work ( haha )

[Chris_Online @Mista_AI]

@Figure_robot At this rate, the robots would hit 1,000,000 packages in about 800hr or around 33days

[Chris_Online @Mista_AI]

@Figure_robot It would be funny if the robot had a hat and a cigarette in his mouth while he was working, thoughts anyone?

[Chris_Online @Mista_AI]

@adcock_brett @Figure_robot Let the Games begin, Wow, what if they were on piece work, who would make more money ? Place your bets everyone!

[Chris_Online @Mista_AI]

@adcock_brett Let's get a pass/fail of the packages sorted and packages per min and hour so it can be evaluated as a Manager would compare to metrics. Would this employee ( robot or human ) make the cut.

[Chris_Online @Mista_AI]

@USAttyPirro I'll take 2 lamps and 2 computers

[Chris_Online @Mista_AI]

Overall this is real progress toward transparency. What’s the top priority for the next update from the community’s perspective?Thanks for open-sourcing the actual production For You algorithm code today.This is a meaningful transparency step: the full candidate pipeline, Phoenix transformer ranking logic, Grox content understanding, and end-to-end inference (run_pipeline.py) are now public. Removing most hand-engineered heuristics in favor of learned transformer scoring is a bold architectural shift.A few real observations / constructive points:The pipeline and inference code are genuinely useful for understanding how the feed works.
Pre-trained mini-Phoenix artifacts (~3GB) are included, which is helpful.
The trained model weights / full checkpoints are still not open, so full reproducibility and independent auditing remain limited.
Would love to see future updates include:More explainability on how individual engagement signals are weighted
Clear changelog + impact notes with every 4-week release (as originally promised)
Thanks to All the X Team for their work !