Economics
The discount is a direct function of trading volume. These worked numbers show what a day of volume buys, and how the pool behaves when demand is higher than it can cover.
From volume to pool
$ORBIX launches on Pons with a custom creator fee. 1.2% of every trade reaches the ORBIX treasury net, and all of it becomes the next day's subsidy pool.
pool = volume × 1.2%| Daily volume | To ORBIX (1.2%) = pool |
|---|---|
| $100,000 | $1,200 |
| $300,000 | $3,600 |
| $1,000,000 | $12,000 |
Fees arrive in WETH and in $ORBIX. The $ORBIX part is sold for ETH, so the whole creator fee can pay for inference.
What the pool buys
At $1,000,000 of daily volume the pool is $12,000. If the average discount across all requests is 50%, the pool covers half of every bill, so it supports about $24,000 of inference at list price per day.
inference supported = pool ÷ average discount
= $12,000 ÷ 0.5
= $24,000 per dayAt list prices on the market today, that is roughly 2.4B Claude Sonnet 5.5 output tokens ($10 per 1M), or about 64B DeepSeek V4.1 Flash output tokens ($0.375 per 1M), every day.
When demand outruns the pool
The pool can't spend more than it has. When projected subsidy demand for the rest of the day exceeds what's left, every tier's discount is scaled down by the same factor:
pool factor = min(1, pool remaining ÷ projected subsidy demand)
discount = tier cap × pool factorExample. The pool has $1,200 left. At current traffic, full-cap discounts would need $1,600 for the rest of the day. The pool factor is 0.75, so a Core holder (60% cap) gets 45% off and a Singularity holder (80% cap) gets 60% off. As traffic eases or the day ends, the factor recovers.
Scaling keeps discounts available all day. A first-come pool would be drained in the morning by whoever starts earliest, and everyone else would pay list price.
Where the money goes
| Flow | Share |
|---|---|
| Creator fee reaching the ORBIX treasury | 1.2% of every trade, net |
| Subsidy pool | 100% of what the treasury receives |
| Kept by ORBIX | 0% |
ORBIX keeps no margin from trading fees and runs no buyback or burn. Every dollar the token earns pays for someone's inference.
Sensitivities
- Volume falls → the pool shrinks → discounts get smaller. This is the main risk and the reason the discount is dynamic.
- Usage grows faster than volume → the pool factor drops below 1 more often. Tier caps stay the same, the realised discount falls.
- Upstream prices fall → the same pool covers more tokens. List prices on the market follow upstream.