Liquid Inference vs OpenRouter

OpenRouter is a hosted router. It sells each model at the list price of the provider and adds a fee when you buy credit. Liquid Inference is an exchange. Providers quote their own prices, and each request starts an auction among the offers that meet your rules.

Both services take the OpenAI and Anthropic formats, so a move changes the base URL, the key and the model names. The facts about OpenRouter come from its own pages, checked on 10 November 2026. The links are in Sources.

Feature by feature

Feature OpenRouter Liquid Inference
Price The provider's list price Each provider quotes its own price
Provider choice A weighted random pick by price, speed and uptime An auction: the lowest charge plus wait cost wins
Fee 5.5% on card top-ups, 5% by crypto, 8% on Business 5% of the charge, or a floor on cheap models. No top-up fee
Cap for each request No. max_price filters only the rate for each million tokens Yes. cap_usd; the provider pays any usage above it
Spend limits Yes. For each key, member and workspace Yes. For each key, workspace, organization and end user
Preview with no charge No Yes. The charge and cap of each offer, and why others are out
Record of each request Yes. GET /generation: provider, cost, tokens and latency Yes. A receipt: rates, fee, cap, charge split and the offers left out
Credit for a failure after output No. Only an empty response is free Yes. A full credit
No retention by default No. Set zdr: true to turn it on Yes. Only offers that keep no prompt compete
Open to sellers Yes. By application; OpenRouter staff edit the listing Yes. By application; sellers post their own offers
Own cloud keys Yes. Free to $25,000 a month at list price, then 5% Partly. Beta. The exchange fee at the list price
Reserved capacity Partly. An Enterprise add-on for capacity that you buy elsewhere Partly. Beta. A monthly pool from one provider

What the exchange does that OpenRouter does not

When OpenRouter is the better choice

Switch in one line

 client = OpenAI(
-    base_url="https://openrouter.ai/api/v1",
+    base_url="https://router.inference.ai.exchange/v1",
     api_key=...,
 )

Then change each model to a name that GET /v1/models lists, and move the provider object to exchange. Move from OpenRouter has each step and the table of settings.

Questions

Does the exchange read the OpenRouter provider object?

Yes. It maps only, ignore, order, max_price, data_collection: "deny", zdr: true and sort: "price". It refuses a setting that it cannot apply exactly, and does not ignore it. See Purchasing rules.

Can a request switch provider in the middle of an answer?

No. The exchange can try the next offer only before the first output. After that, it never switches provider. You never get one answer from two providers.

What changed in OpenRouter's routing on 28 October 2026?

OpenRouter replaced its price-first default with a routing cost for each provider. The routing cost includes price, throughput, latency and reliability. OpenRouter reported 6% higher cost in its own test. To go back to price first, set provider.sort to price.

Who owns OpenRouter?

On 19 August 2026, Stripe said that it agreed to acquire OpenRouter. On 10 November 2026, we found no statement that the sale is complete.

Sources

Checked on 10 November 2026.