Skip to content
NewTry RouterPlus with free creditsTry playground →
RouterPlus
Book a demo

AI gateway for every LLM

One hosted key in front of 43 models from 16 providers. OpenAI- and Anthropic-compatible, with failover, spend caps and the cost of every request.

Open in consoleI need inference
  • 43Models
  • 16Providers
  • 2API formats: OpenAI and Anthropic
  • 0%Inference fee with your own provider key
Providers
  • DeepInfra
  • Inference.net
  • StreamLake
  • Doubleword
  • Nebius
  • Phala
  • Novita AI
  • Chutes
  • Sail Research
  • Wafer
  • Morph
  • RunInfra
  • Munito AI
  • Particle AI
  • Mistral AI
  • Paper Instruments

What the gateway does

  • Two API formats, one key

    OpenAI- and Anthropic-compatible endpoints. Every chat model answers on both, and the gateway translates between them.

    Docs

  • Failover before the first token

    A provider that fails before any output is skipped with zero backoff. Once output starts, your request stays with that provider.

    Docs

  • The cost on every response

    Every billed response carries usage.cost in USD, including attempts that failed over first. You can rebuild your bill from the wire.

    Docs

  • Spend caps and rate limits

    Monthly spend caps per key and for the whole org, plus requests, tokens per minute and concurrency limits you set per key.

    Docs

  • Bring your own key

    Connect OpenAI, Anthropic, Azure or Bedrock accounts. The provider bills you directly and RouterPlus charges a 0% inference fee.

    Docs

  • Routing you can pin and explain

    Order or pin providers per request, save a routing policy per key, and ask POST /v1/route for the plan without sending the request.

    Docs

Quickstart

OpenAI-compatible
curl https://api.routerplus.com/v1/chat/completions \
  -H "Authorization: Bearer $ROUTERPLUS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v4.1-flash",
    "messages": [{ "role": "user", "content": "Hello" }]
  }'

Endpoints

EndpointWhat it does
POST /v1/chat/completionsOpenAI-compatible chat completions, streamed or not
POST /v1/messagesAnthropic-compatible messages, streamed or not
POST /v1/responsesOpenAI Responses API for OpenAI's models
GET /v1/modelsModels your key can call, in OpenAI or Anthropic list shape
POST /v1/images/generationsOpenAI Images API for image models
GET /v1/usageBalance and recent requests with their cost
GET /v1/generationEvery attempt behind one request: tokens, provider, cost, outcome
POST /v1/routeHow a request would route, without sending it
GET /v1/limitsThe rate limits that apply to your key

Host: https://api.routerplus.com. Same key on every endpoint.

Compare AI gateways

Reserved capacity

Tell us the model, throughput and region you need. RouterPlus shops the requirement across its provider network and replies with committed capacity, a better quote or a backup provider.

Send a capacity request

FAQ

What is an AI gateway?

An AI gateway sits between your app and the model providers. Your code calls one API with one key, and the gateway picks a provider, fails over when one is down, enforces your limits and records what each request cost.

Is an LLM gateway the same thing?

Yes. LLM gateway is the same idea named for language models. RouterPlus serves chat models on its OpenAI- and Anthropic-compatible endpoints and image models on the OpenAI Images endpoint.

Do I have to host the gateway?

No. RouterPlus is hosted at https://api.routerplus.com/v1. There is nothing to deploy or upgrade: sign up in the console, add credits and call any of 43 models from 16 providers.

What happens when a provider fails?

If a provider fails before any output reaches you, the gateway tries the next one with no backoff, and the x-tm-attempts header counts the tries. After output starts, the request stays with that provider. Content-policy refusals are never sent to another provider.

Which SDKs and tools work with it?

OpenAI SDKs, Codex, Cursor, Cline, Kilo Code, OpenCode and anything that speaks the OpenAI format use the OpenAI-compatible endpoint. Anthropic SDKs and Claude Code use the Anthropic-compatible endpoint. The same key works on both.

How is the gateway priced?

Pricing is pass-through: you pay each provider's listed per-token rate from prepaid credits, and every billed response states its exact cost. With your own OpenAI, Anthropic, Azure or Bedrock key, the provider bills you and RouterPlus charges a 0% inference fee.

Does the gateway store my prompts?

Billing and analytics keep metadata only. Requests to shared endpoints go to an encrypted request log for up to 30 days, for debugging; dedicated endpoints and organizations that opt out are never logged. The provider serving your request applies its own data policy.

One key for every provider

Open in consoleI need inference