The real skill with open models isn't running them. It's routing
Small model answers first and says how sure it is. Only unsure tasks go to the frontier model. A test, not the model, decides if it's done
Cheap by default. Expensive by exception
Flight cancelled. 187 people ahead of you in line
You stay in your seat. Your AI phones the airline, waits on hold, rebooks you
That's the viral claim. The calling feature is real: Grok Bot can dial for you now
The flight? Never named
Would you trust it with your trip?
4
15
459
Viral post:
Grok's announcement: teapawt.net/grok/status/2118138308…
holy f*ck... Elon Musk's team just made airport rebooking lines useless
my flight got cancelled and 187 people were ahead of me at the desk. my Grok Bot rebooked me before the line moved 17 people
Grok Bot makes phone calls now. it sits on hold, talks to the agent and says up front it's an AI
what it did while I stood in line:
→ sat through 11 minutes of hold music
→ got me on the 7:40 AM flight tomorrow
→ the airline cancelled, so it asked for a hotel and meals. got both
→ "aisle seat is $49 extra" it didn't pay. it asked me. I took the free middle seat
→ texted my wife the new landing time (signed Grok Bot) and set my alarm
14 minutes on the phone. I walked out of the line, 168 people still in it
all it took was 3 rules I sent it once:
1. on every call, say you're an AI calling for me
2. ask for everything I'm owed, nothing made up
3. anything that costs money: ask me first
the Grok Bot Research Desk setup is in my article. read it below ↓
Readers added context they thought people might want to know
This post presents a staged demo video of a hypothetical Grok Bot rebooking scenario on the fictional airline Bluepeak Air as a real personal event.
x.com/gdalt3/status/…
en.wikipedia.org/wiki/List_of_a…
teslanorth.com/2026/11/04/gro…
x.com/bot/status/211…
25
~12,000 flights cancelled. 3,500 planes in the air. A nationwide ground stop
Hack? AI? The FAA’s early answer: a data flood, 1,000 connection attempts every 4–5 seconds, from a site getting a new router
Polymarket: hack 3%, AI 2%
Boring cause. Huge chaos
4
21
352
Market: polymarket.com/event/what-te…
2
29
This text-to-speech model has never heard a real human voice
It learned from 19 hours of speech made by a bigger AI. Result: 4.5M parameters, 18 MB, ~70x real time on a CPU
Total cost: $9.56. Weights are open
Send this to anyone who thinks AI needs a data center
I trained a text-to-speech model that has never heard a real human voice.
4.5M parameters. 18 MB. Runs on a CPU.
Meet tiny-voca
8
22
634
Source: author's thread
Code & docs: github.com/n4ze3m/tiny-voca
I trained a text-to-speech model that has never heard a real human voice.
4.5M parameters. 18 MB. Runs on a CPU.
Meet tiny-voca
1
2
57
Your AI agent says “it works.” Now it has to show the video
Expo launched cloud simulators for agents: the agent runs your app on a cloud iPhone, taps through its own change, and brings back a screen recording as proof
No Mac needed
Merge a PR with video, not “trust me”?
1
15
468
One photo in. A whole 3D world out
World Labs just opened its Atlas model to everyone. Give it a single image and it builds a scene you can fly the camera through, including parts the photo never showed. It imagines those
Send this to the friend who thinks a photo is flat
We’re opening up access to Atlas to everyone today.
We’ve built apps and API endpoints to help you use Atlas any way you can imagine: bring your ideas for building a 3D world and come away with keyframes, video clips, or 3D Gaussian splats.
2
19
716
Elon may merge SpaceX and Tesla into a ~$3.4T giant. Traders say 73%
Bloomberg: a deal could land as soon as December. On Polymarket it’s only 31% by Dec 31, but 73% by June 2027
Neither company has commented
Send this to the friend who thinks Elon’s empire stays split
3
17
313
Google just made an AI that doesn't write a single word
Show it one photo, ask 1,000 yes/no questions. It answers all of them at once, with a confidence score for each, in about 0.2 seconds
Send this to the friend who still waits for AI to "finish typing."
Today we’re launching the Gemini Decisions API, powered by our new Gemini Decision 1 model in @GoogleAIStudio.
Instead of generating text token by token, it’s built for closed-set choices scoring up to 1,000 questions across text, images, audio, video + PDFs in a single pass. A few details:
- 3 decision types: yes/no, multiple choice (up to 255 options) + rubric scoring
- $0.10 / 1M input tokens with $0 billed for output tokens (~5x cheaper than Flash-Lite)
- leads on several benchmark evals with lowest calibration error
Google teams are already building with it, below is a look at Stitch Rapid designing a full product flow in seconds (1x speed)
1
10
444
Windows 11 now runs on a foldable iPhone
Ammaar Reshi says Gemini 4 Carbon helped him build the engine. Unfold the screen: a pocket PC with coding agents
The catch: you need a Mac, Xcode and your own Windows image
Send this to the friend who still carries a laptop
I used Gemini 4 Carbon to get Windows 11 running on my iPhone Duo! 💻
Unfold the screen and it's a pocket PC, including Antigravity running coding agents on my phone.
This is the actual desktop ARM OS running locally with touch controls.
Demo + repo below!
4
13
580









