Is there an LLM API marketplace where sellers compete on price?
Yes. UberLLM is an LLM API marketplace where sellers compete on price: each seller sets its own price for a model as a discount off the model’s list price, and cheaper offers that are working get more of the requests. On November 19, 2026, 228 chat models were taking requests through one OpenAI-compatible API. 244 sellers on the market can serve DeepSeek V4.1 Flash, the most used. UberLLM adds no fee to requests; adding credit by card costs 5.5% + $0.50. A request that fails is not billed.
In a seller market, several sellers offer the same model, and each one sets its own price. A buyer uses one API key and one address. Each request goes to an offer for that model that is working right now, and the buyer pays what that offer charged.
So the price of a model is not one published number. It is the lowest working offer at that moment, and it moves when sellers change their prices or get busy.
How the UberLLM market works
A seller lists an offer: one model, served through the seller’s own account with a provider (an API key at that provider’s address), at a price the seller picks. Sellers either list directly on UberLLM or sell on a partner inference market that UberLLM connects to.
Each seller sets its price as a discount off the model’s list price (the price you would pay the model maker directly).
When you send a request, it goes to an offer for that model that is working right now, and cheaper offers get a larger share of requests. If that provider fails before it starts answering, the request is retried with the next one.
Health is measured on real requests. An offer whose requests fail is taken out of routing for a while, and longer each time it fails again.
Each request is billed what its provider charged, from prepaid credit. UberLLM adds no fee to requests; adding credit by card costs 5.5% + $0.50. Requests that fail before the provider starts answering are not billed.
UberLLM does not run the models. Listing is free, and UberLLM takes no fee from seller sales.
The whole description, with who the sellers are and what they must agree to: How UberLLM works. To list capacity: Sell capacity.
Seller market or router: where the price comes from
From the code, a router (also called an LLM gateway) and a seller market look the same: one key, one base URL, and the model named in each request. UberLLM is both. What differs is where the price of a model comes from.
A router with fixed provider prices: each provider publishes one price per model, and a request is billed at the price of the provider that served it. The price changes when a provider changes its price list.
A seller market: several sellers offer the same model and each sets its own price, so sellers compete on price. The price shown is the lowest offer right now, and a request can cost more when the cheapest sellers are busy.
On a seller market the buyer can set a price too. In a chat request, provider.max_price is the most it may cost per million tokens: the request only goes to offers at or under it and is never billed above it.
Check it yourself
The current lowest price of every model is public, with no API key and no account. Prices are read again every 30 seconds.
A request with a price ceiling, here the list price of Google: Gemini 4 Hydrogen per million tokens. If no seller is within the ceiling, the request is refused with 400 max_price_exceeded and nothing is charged.
Prices move. The price shown is the lowest offer right now, and a request can cost more when the cheapest sellers are busy. For a hard limit, send provider.max_price.
A third party serves the request. The provider that serves a request receives what you sent and handles it under its own terms. UberLLM saves each request and its answer for 90 days, then deletes them, and never uses them to train models. You can turn this saving off for any API key.
The model is tested, not proven. Sellers who list on UberLLM can list only through a provider on its approved list, and the provider’s model must be the model listed. Every offer is tested and reviewed by a person before any buyer reaches it, and keeps being tested on live requests. But no software check can prove which exact copy of a model wrote an answer; only hardware attestation can, and UberLLM does not have it.
Precision. Some providers serve smaller 8-bit or 4-bit copies of an open model, and those can pass the tests. Sellers must state the precision they serve, and a false statement gets them banned.
The partner market has its own rules. Requests for Claude, GPT and Gemini go to sellers on a partner market. It requires each seller to be authorized to resell that access under the provider’s terms and checks that each offer answers correctly; it does not prove which model produced an answer.
UberLLM cannot see a seller’s own costs or contracts. The rules do not change with the price: sellers must own or be authorized to resell the access they list, and they are paid only for requests they actually served, after refunds and chargebacks.
What the tests catch, what they cannot, and what happens when a seller cheats: Model checks.
How is a seller market different from an LLM router?
Both give one key and one API for many models. A router with fixed provider prices bills each request at the published price of the provider that served it. On a seller market several sellers offer the same model at prices they set themselves, so the price of a model comes from competing offers and moves during the day.
Who are the sellers?
A seller lists a model it can serve through its own account with a provider and sets a price. Sellers either list directly on UberLLM or sell on a partner inference market that UberLLM connects to. They must have the right to resell that capacity under their own provider’s terms.
Do I get the model I asked for?
Sellers can list only through a provider on UberLLM’s approved list, and the provider’s model must be the model listed. Offers are tested before any buyer reaches them and keep being tested on live requests; what the tests cannot prove is stated at uberllm.dev/verification.
How do I pay on UberLLM?
Prepaid credit by card or PayPal, from $5, with no subscription. A request that fails is not billed, and every request is listed with its cost in Dashboard → Activity.