Sell your surplus inference
to the highest-value demand.
List an OpenAI-compatible endpoint on the UberLLM order book. When you are the cheapest healthy offer for a model, buyer traffic routes to you automatically — and settles to a USD balance you can withdraw.
No listing fee · Passive health checks · USD payouts · Delist anytime
Live on the marketplace · last 24h
How selling works
Four steps from spare capacity to a USD balance. No sales calls, no contracts.
List an endpoint
Point us at any OpenAI-compatible /v1 endpoint — your own GPUs, a hosted deployment, or resale-permitted credits. Add the key once; it is encrypted and never exposed to buyers.
Price your offer
Set an absolute $/MTok for input and output, or a multiplier of the live reference (e.g. 0.9× to always sit just under list). Change it anytime; the book re-sorts instantly.
Win real traffic
You receive requests whenever you are the cheapest healthy offer for a model. Health is measured passively from real traffic — we never burn your quota with synthetic pings.
Get paid in USD
Every served request settles to your payable balance at the price it routed at. Withdraw to your payout method — no tokens, no lockups, no minimums beyond payout rails.
Price it your way
Two pricing modes, switchable per offer at any time.
Fixed $/MTok
Name your own input and output price per million tokens. Best when your unit economics are known and you want a hard floor.
A multiple of the reference
Track the live reference price at, say, 0.9× so you always sit just under list without re-pricing by hand as the market moves.
Set a daily cap to bound spend against your key, and a region so latency-sensitive buyers can prefer you. Full controls live in the seller console.
Earnings, worked out
Illustrative — your numbers depend on your price, volume and cost basis.
| Reference price (in / out) | $0.60 / $0.80 per MTok |
| Your multiplier | 0.9× |
| Your price (in / out) | $0.54 / $0.72 per MTok |
| Volume served | 50M input + 15M output / day |
| Gross / day | $37.80 |
| Gross / month | $1,134.00 |
Settlements accrue at the exact price each request routed at. UberLLM never marks up your tokens — buyers pay your price, you keep it.
An offer is just an endpoint and a price
This is the whole listing payload. Create it in the console, or via the seller API.
{
"model_id": "meta-llama/llama-3.3-70b-instruct",
"endpoint_url": "https://your-host.example.com/v1",
"pricing_mode": "multiplier",
"multiplier": 0.9,
"daily_cap_usd": 250,
"region": "us-east"
}List only capacity you are allowed to resell
Sell inference you own or are licensed to resell — your own GPUs, a self-hosted open-weights deployment, or credits under a plan that permits resale.
- Do not resell access to a frontier-lab API key (OpenAI, Anthropic, Google and similar) — their terms prohibit sublicensing, and such offers will be delisted.
- Do not list capacity you cannot reliably serve; sustained health failures pause an offer automatically.
- You are responsible for the licensing and lawful operation of the endpoint you point us at.
Put your surplus on the book
Create an account, open the seller console, and list your first offer in a couple of minutes.