Liquid Inference vs Portkey
Portkey is a gateway in front of your own provider accounts. It adds logs, guardrails and routing rules, and you pay each provider yourself. Liquid Inference is an exchange. Providers quote their own prices, and each request starts an auction among the offers that meet your rules. You pay every provider from one balance.
Both services take the OpenAI and Anthropic formats. The facts about Portkey come from its own pages, checked on 10 November 2026. The links are in Sources.
Feature by feature
| Feature | Portkey | Liquid Inference |
|---|---|---|
| What you buy | Gateway software. You keep an account at each provider | Inference. One key and one balance for each provider |
| Price | Your contract with each provider | Each provider quotes its own price |
| Provider choice | A config that you write: fallbacks, retries, splits and conditions | An auction: the lowest charge plus wait cost wins |
| Fee | A plan: free to 10,000 logs a month, $49 for 100,000, or Enterprise | 5% of the charge, or a floor on cheap models. No plan fee |
| Cap for each request | No | Yes. cap_usd; the provider pays any usage above it |
| Spend limits | Partly. Budgets on Enterprise and some Pro plans | 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. A log with a cost that Portkey calculates | Yes. A receipt at the rates of the offer |
| No prompt logs by default | No. Kept 3 to 30 days on public plans. You can turn them off for a request | Yes. The exchange stores no prompt or answer |
| Open to sellers | No. Only the providers where you have an account | Yes. By application; sellers post their own offers |
| Self-host | Yes. Open source, under the MIT licence | No |
What the exchange does that Portkey does not
- One balance for every provider. You need no account, contract or key at each provider.
- Prices from competition. Each provider quotes its own price for each model, and the auction buys the lowest cost. See Price and wait.
- A hard cap for each request. Send
exchange.cap_usd, or thex-liquid-cap-usdheader. See Purchasing rules. - A dry run. See which offers compete and what the request can cost before you buy. See Purchasing rules.
- No prompt logs. See Data handling.
When Portkey is the better choice
- You have contracts with your providers, and you want one control point in front of them.
- You must run the gateway in your own network.
- You need a large set of guardrails on requests and answers.
Switch in one line
client = OpenAI(
- base_url="https://api.portkey.ai/v1",
+ base_url="https://router.inference.ai.exchange/v1",
api_key=...,
)
Then remove the Portkey headers, and change each model to a name that GET /v1/models lists. Move from Portkey shows how to replace a Portkey config with an exchange object or a preset.
Questions
Can I keep my own provider keys?
Yes. Connect a cloud key on the exchange. Your key then competes in the auction, and the cloud bills you. The exchange charges only its fee. This is in beta. See Bring your own key (BYOK).
Does the exchange have guardrails?
Yes. A workspace guardrail sets offer limits, budgets, rate limits, and checks for prompt injection and sensitive data. No request can widen it. See Guardrails.
Who owns Portkey?
Palo Alto Networks completed its acquisition of Portkey on 29 May 2026. Portkey now also sells the gateway as Prisma AIRS AI Gateway.
Sources
Checked on 10 November 2026.
- Product and keys: Model catalog, Portkey
- Routing: AI Gateway
- Plans and logs: Pricing, Logs
- Budgets: Budget limits
- Formats: Universal API
- Open source: Portkey gateway on GitHub, Gateway 2.0
- Ownership: Palo Alto Networks press release