We're opening ZERO, a $0 Developer Alpha for coding intelligence, backed by a shared 100T-token daily pool.
There's an optimization developers using AI have learned almost unconsciously:
Think about the tokens.
How large is the repository?
How much context should I send?
Should I let the agent keep going?
How expensive will this task become?
What if we removed that constraint?
That's the experiment behind Project Zero.
100 trillion tokens per day
Project Zero starts with a shared developer pool of:
100,000,000,000,000 tokens every day.
That's 100T tokens/day.
The pool resets at 00:00 UTC, and access during the Developer Alpha costs developers $0.
At the center of the experiment is ZERO.
Its model identity is:
tessen/project-zero
The idea is not to create another AI chat interface.
We want ZERO available where developers actually build software.
Our first major focus is OpenCode.
The desired experience is deliberately simple:
Open coding environment
↓
Select ZERO
↓
Open repository
↓
Build
What should developers do with 100T tokens?
Use them.
Not on toy prompts.
Point ZERO at real software.
Give it large repositories.
Ask it to understand unfamiliar systems.
Debug difficult problems.
Refactor architecture.
Build features.
Run migrations.
Let coding agents work through long tasks.
And find where the system breaks.
One of the first questions we want Project Zero to answer is:
Can developers actually consume 100 trillion tokens in 24 hours?
At midnight UTC, the experiment starts again.
There's another side to Project Zero
Giving developers free intelligence creates demand for compute.
So we're also building Tessen Node.
Node allows compatible machines to participate in Tessen's distributed compute network and contribute verified capacity.
That gives Project Zero a second experiment.
Can we create this flywheel?
MORE DEVELOPERS
↓
MORE DEMAND
↓
MORE NODES
↓
MORE VERIFIED COMPUTE
↓
MORE INTELLIGENCE
↓
MORE DEVELOPERS
That's where the Project Zero principle comes from:
MORE DEVELOPERS. MORE NODES. MORE INTELLIGENCE.
Building it in public
We're not going to manufacture numbers around this experiment.
If Project Zero has 100 developers, we'll have 100 developers.
If developers consume 1T tokens, we'll report 1T.
If Nodes contribute capacity, we'll measure verified capacity.
And if something doesn't work, that's part of the experiment too.
The interesting numbers should come from what developers actually do.
Project Zero Developer Alpha
The starting proposition is:
ZERO
100T TOKENS / DAY
RESET 00:00 UTC
$0 DEVELOPER ALPHA
tessen/project-zero
The ZERO Pool is being prepared now.
When it opens, we're looking for developers willing to push it against real software and find its limits.
Follow the experiment
Project Zero:
https://github.com/TessenLab/zero
Tessen:
Tessen Platform:
Run a Tessen Node:
If you work with coding agents, large repositories, AI-assisted software engineering, inference infrastructure or distributed compute, we'd particularly like you involved.
What happens when coding intelligence costs $0?
We're going to find out.
THE ZERO POOL IS COMING.
Top comments (0)