Quickstart — Migrating from a Gateway

Already calling OpenAI, OpenRouter, Vercel, Cloudflare, Anthropic or Gemini? Migrating to FreeRouter is two edits: change the base URL and swap the key. No SDK changes, no request rewrites.

Starting from scratch? This guide assumes you already call an LLM gateway. New project, no gateway yet? Follow Starting fresh instead.

The two changes

First, bring your current gateway's API key with you: paste it under Providers and create a FreeRouter key routed to it, with the key's API shape set to the format your client speaks — the three dashboard steps are in Getting Started. Then change two values in your code:

You call today	Base URL before	Base URL after	Key before → after	Key shape
OpenAI	`https://api.openai.com/v1`	`https://api.freerouter.com/v1`	`sk-…` → `fr_live_…`	`openai`
OpenRouter	`https://openrouter.ai/api/v1`	`https://api.freerouter.com/v1`	`sk-or-…` → `fr_live_…`	`openrouter` (body forwarded verbatim)
Vercel AI Gateway	`https://ai-gateway.vercel.sh/v1`	`https://api.freerouter.com/v1`	your gateway's API key → `fr_live_…`	`openai`
Cloudflare AI Gateway	`https://gateway.ai.cloudflare.com/v1/{account}/{gateway}/openai`	`https://api.freerouter.com/v1`	API token → `fr_live_…`	`openai`
Anthropic	`https://api.anthropic.com`	`https://api.freerouter.com` (no `/v1`)	`sk-ant-…` → `fr_live_…` (as `x-api-key`)	`anthropic`
Gemini	`https://generativelanguage.googleapis.com/v1beta`	`https://api.freerouter.com/v1` — `/models/{model}:generateContent` only	`AIza…` → `fr_live_…` (as `x-goog-api-key`)	`google`

The account id, gateway name or project your old URL carried is gone from your code: FreeRouter stores those with the provider key. Model ids are provider/model everywhere, e.g. openai/gpt-4o-mini.

That's the whole migration Endpoints, request bodies, streaming and tool calls stay identical. What does change behind the proxy — added headers, the 10 MB body limit, the few endpoints not served — is on one page: What's different. Per-SDK before/after snippets are in Use your SDK.

Migration assistant — generate an agent prompt

Switching from another gateway? Use the wizard below to construct a prompt you can paste into your coding agent. The prompt asks the agent to study the FreeRouter docs and your code, then come back with a migration plan for your review — not to start rewriting code.

Step 1 — Which gateway do you use today?

Step 2 — How does your code call it? (optional)

Step 3 — Paste this prompt into your agent

Raw HTTP with curl

The request is byte-for-byte what you send the gateway today; only the host and the key differ. "stream": true returns server-sent events exactly as before:

curl https://api.freerouter.com/v1/chat/completions \
  -H "Authorization: Bearer $FREEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -N \
  -d '{
    "model": "openai/gpt-4o-mini",
    "stream": true,
    "messages": [{ "role": "user", "content": "Count to 3." }]
  }'

In your SDK

OpenAI Python and Node, Anthropic Python and Node, Claude Code, Vercel AI SDK, LangChain, LiteLLM, the Responses API, Gemini, or just two environment variables — each with a before/after and a streaming variant — are in Use your SDK.

Verify

The response carries X-FreeRouter-Provider naming the gateway that answered; GET /v1/models lists what the key can reach.
A 401, 402 or 404 on the first call has a one-line fix in Troubleshooting — the Anthropic SDK's /v1-on-the-base-URL mistake and the unfunded-account 402 are the usual two.

Choosing a model

Pass a model string in the model field. FreeRouter uses provider/model style identifiers so a request is unambiguous across gateways — for example openai/gpt-4o-mini or anthropic/claude-haiku-5.5. Your routing rule then decides which gateway actually serves that model. List what's available with the /v1/models endpoint.

Different API shape? If your service speaks the OpenRouter, Anthropic, or Google format instead of OpenAI, set the key's API shape to match — the base URL swap is the same.

Next Use your SDK