We are ready to build worlds.
Currently fire-testing the new incentive mechanism that will get miners to build our first ever generative world model, capturing real physics, long-term consistency, and true interactivity.
This is the unreal next chapter for vision AI: from understanding scenes to generating entire, livable realities.
This 👇 is a sneak peek of the subnet’s console, where users will be able to create worlds from a single prompt and very soon explore them right in their browsers.
This will also 100x our ability to train better vision models by providing unlimited simulated scenarios.
44 we go.
Counting penguins from a rocking boat in Antarctica. Cooking at @webuildscore 🐧
Nothing in this shot holds still. The camera’s on a moving ship, the iceberg drifts, and half the penguins are mid-jump. The count stayed steady from the first frame to the last.
First week of November on SN44.
Person 90.6% / goal 90%
Vehicle 90.2% / goal 90%
Worksite PPE 90.9% / goal 90%
Room dirtiness 85.8% / goal 90%
Beverage 71.1% / goal 90%
Cricket delivery 73.0% / goal 85%
Three already past their goal, the rest get pushed every hour.
📹 A billion cameras. One AI network.
@MaxSebti of @webuildscore tells @2084Richardson at @ExploitSummit why he sees Bittensor as “Bitcoin for AI” and a network state with its own growing GDP.
If that sounds too big, that’s exactly why you should hear the rest.
Had a solid week 44 IRL last week.
Doing this over the next 48 hours to celebrate it 👇
1️⃣ SN44 emissions → 44% miners, 44% validators, 12% owner.
2️⃣ Convicting 44% of the owner’s wallet.
3️⃣ Take a guess, you're always asking for it.
Forever Forty-Four.
Counting 500+ shipping containers from a drone clip.
Freight runs on knowing exactly what is in the yard. One misplaced box means idling trucks and daily fines.
I wanted to see if the new build I'm testing at @webuildscore could handle the density without dropping targets between frames.
It kept the count live and didn't miss the small containers tucked into the middle of the stacks.
Old world:
→ Computer vision companies siloed their own data and tools, spending months or years solving niche problems. Progress was fragmented and negative-sum.
New world (SN44):
→ Detectors and segmentation models are improved fully open-source, building the largest and most efficient computer vision model library in the world, available instantly, via app or on Hugging Face.
The issue was never a lack of data, talent, or technology.
It was misaligned incentives. Once those incentives align around open collaboration, the entire vision AI ecosystem compounds instead of competing.
$TAO A few days ago I said the use cases were everything we can see in the viewable universe.
The answer is everything we can view. Probably leading to almost everything we can’t see. @webuildscore
Had a solid week 44 IRL last week.
Doing this over the next 48 hours to celebrate it 👇
1️⃣ SN44 emissions → 44% miners, 44% validators, 12% owner.
2️⃣ Convicting 44% of the owner’s wallet.
3️⃣ Take a guess, you're always asking for it.
Forever Forty-Four.
Beekeepers judge a hive by its entrance. Busy and steady means the colony is healthy. A sudden drop or a frantic spike usually means trouble.
So I tried what I’ve been testing at @webuildscore on it. One class, a few labels, and it counts. Early build. I fixed one screw it kept calling a bee, everything else is straight from the first run.
Gotta keep improving.
New on SN44, room dirtiness.
Spills, litter and bins, live since the weekend. It’s at 85.5% against a 90% target, so a bit more work to do.
Like every new detector, we’ll keep making it harder over time.
October on SN44, in a few lines.
-More public tracking tasks running on the subnet
-3 new public elements out
-A big chunk of example data from the cricket private track is now public
-A lot of quiet work on security and scale
-We can spin up new tasks faster now, with plenty more coming
And following IBC, one of the largest real-time graphics software companies is now testing a Studio endpoint, running on the football model built earlier this year under our previous incentive mechanism.
Meanwhile, we’re collecting feedback from Studio users for what comes next.
From here the pace only picks up, with way more tasks every month and one goal, making computer vision a commodity.
Early build of what I’m testing at @webuildscore. Cleaner, smoother, and with live count.
The old version was a wall of tiny labels and overlapping boxes, hard to tell what it was actually seeing. This one is easier to read at a glance and keeps up with the line as it moves.
Still a few rough edges, and still a lot more coming.