Quern

Browser compute on Solana. Your GPU searches millions of hashes, the server checks the answer with one. Only work it can verify is paid.

[Quern @Quern_gpu]

$QUERN is live.

CA : BoUa7eK8UuzChZPjUU9Aqatc7qngg4QFXyEnukzLpump

Browser compute on Solana. Your GPU searches for a SHA-256 preimage; the server
checks the answer with one hash. Only work it can verify is paid.

No AI training. No customer renting compute. Nothing your browser claims about
itself is priced.

https://t.co/F6Kn61YvWv

[Quern @Quern_gpu]

Difficulty adapts to how quickly your node actually solves tasks. A fast GPU
receives harder tasks, not more of them.

One session climbed from 20 bits to 28 from about a million hashes per task to
268 million.

Reward scales with 2^difficulty, because expected work does

[Quern @Quern_gpu]

Locked all supply :

https://t.co/JUNHucZj4M

[Quern @Quern_gpu]

A shader computing the wrong hash still returns numbers. Still looks busy. It
would simply never solve anything, and would look like bad luck for hours so a
node is not allowed online until the CPU confirms an answer its GPU found.

And nothing it reports afterwards is priced: not its speed, not its hardware, not
how long it says a task took.

[Quern @Quern_gpu]

The network so far, counted from its own database:

tasks verified 45
tasks rejected 0
hashes 7,820,279,808
issued 298.32 QUERN

Every one of those was checked server-side before it was credited.

[Quern @Quern_gpu]

If you are looking at a compute network, ask it one thing:

**what does verification cost you?**

If the server has to recompute the workload to check it, the server is doing all
the work it is paying for, and the network has harvested nothing.

QUERN's answer is one hash.

[Quern @Quern_gpu]

Finding an input whose SHA-256 starts with 28 zero bits takes about 268 million
attempts.

Checking one takes exactly one hash.

That gap is the entire design of QUERN. The network verifies every result it pays
for without redoing the work.