
The world’s largest camera infrastructure for autonomous driving, and physical AI. Powered by @Solana.
Joined April 2022
- Tweets3.5K
- Following124
- Followers133K
- Likes7.4K
Pinned Tweet
NEW AI CUSTOMER BUYS 3,000 HOURS OF NATIX DRIVING DATA 🚗
We've signed a six-figure deal with an AI lab pioneering general world models. Real demand for DePIN on @solana.
Not a research collab. Not a free dataset. A paying customer.
Passing a truck on a curving ramp with a wall on the other side, the driver speeds up from 70 to 110 km/h in under 2 seconds 🚛
Not every reaction is a brake. Sometimes it's the accelerator.
NATIX telemetry records both, so models learn from what drivers actually do.
1/ Demand for NATIX data is growing fast, and we're in talks for much bigger contracts 📈
So from December 1, Desktop App drivers in the top tiers earn USDC on top of their NATIX rewards, and all Desktop App drivers earn 4X points in December.
Here's what changes and why 👇
6/ A driver doing 4,000 km a month:
Today: Roadmaster
Device uploads: Asphalt Pilot
Desktop App: Road Captain + 50 USDC
That's 42,000 NATIX + 50 USDC more in December than today, for the same driving.
1
4
240
7/ The USDC rewards run as a three-month test, December to February. If they bring more and better data, we'll double down.
Not on the Desktop App yet? Switch before December 1: three steps, your points carry over, and your VX360 keeps collecting 👇
natix.network/blog/usdc-rewa…
5
234
Computer vision is any AI that understands images, from roads to MRI scans. Machine vision is computer vision built into a machine.
Physical AI goes further and acts, like a car that drives itself. It learns from what real drivers do, and NATIX records that 👇
natix.network/blog/machine-v…
Top row: real driving. Bottom row: what Valeo's VATIX world model generated from just the first frame 🚗
Small models get the road right but soon lose the cars. The 9B model keeps them moving for longer.
It was trained from scratch on 5,500 hours of NATIX footage.
1/ VX360 drivers outside the U.S., the Desktop App Open Beta is live 🚀
Keep collecting in your Tesla as usual, then upload from your computer, up to 5X faster than uploading from the device.
3/ Switching takes three steps:
1. Download the app and log in with your VX360 account
2. Update your VX360 on Wi-Fi (2 to 3 hours)
3. Plug it into your computer to upload, then put it back in your Tesla
Upload footage first: migration wipes the device and is one-way.
4/ Migrate now and earn 3X points on eligible kilometers driven through December 31 ⚡
You can also work toward Apex Driver, a new top tier with a 340,000 In-App NATIX reward.
Everything you need to switch 👇
natix.network/blog/vx360-des…
3
312
NATIX Network retweeted
Rare events like this are why we collect from all of a Tesla’s cameras, not just the front one.
Catch it once in the real world, and a world model can learn it, replay it and vary it until every driving stack has seen it a thousand times.
My brother’s friend drives a Tesla. This is 3 videos of the same accident. The Tesla, on self drive, sees the approaching car coming from behind at excessive speed and moves out of the way. Driver of speeding car was killed. Amazing technology @elonmusk.
An SUV starts pulling out of a side road as we approach. The driver brakes hard 🛑
A VLM would describe that moment in words. A VLA would brake, like the driver did.
NATIX uses VLMs to find moments like this one in over 207,000 hours of driving.
A U.S. study recorded over 1.4 million hours of driving and found about 4,500 crashes and near-crashes 🔎
That's about one every 300 hours. Nobody can watch that much video.
NATIX finds them with telemetry, metadata and VLM search. How it works 👇
natix.network/blog/scenario-…
Most video world models predict one frame after another, so small mistakes add up over a long drive.
Orbis 2 plans the scene first, then draws the frames: 4x more stable and 35% more accurate than NVIDIA Cosmos 2.5.
Over 40% of its training data came from NATIX.




