FreeRouter is a mediation layer for AI inference. Integrate once and route every request across OpenRouter, Vercel, Cloudflare and more — swap gateways any time, with no code changes.
Bring your own keys · OpenAI-compatible · No migration project
# Change one line. Keep your code. client = OpenAI( base_url="https://api.freerouter.com/v1", api_key="fr_live_...", ) resp = client.chat.completions.create( model="openai/gpt-4o-mini", messages=[{"role": "user", "content": "Hello, FreeRouter!"}], ) # FreeRouter routes it per your rules.
A neutral layer between you and every gateway. If a provider raises prices or drops a model, flip a routing rule — no migration.
One API key, one base URL, one integration. FreeRouter normalizes the request shape and forwards it wherever your rules point.
Bring your own keys. You keep control of your provider relationships and often pay providers directly — cheaper than a gateway's markup.
Drop in the keys for the gateways you already use — OpenRouter, Vercel, Cloudflare and more.
FreeRouter issues a single key you use in your app. Behind the scenes it swaps keys and normalizes formats.
Split traffic by percentage, or set priorities with failover — "use Cloudflare for cheap models, fail over to Vercel if OpenRouter is slow."
Change providers any time from the dashboard. Zero code changes, zero migration projects.
The default shape is OpenAI-compatible /v1/chat/completions — the
format 90% of developers already use. Prefer another? Toggle the shape per key to OpenRouter,
Anthropic, or Google with one click.
BASE_URL at https://api.freerouter.com/v1# Before export OPENAI_BASE_URL="https://openrouter.ai/api/v1" export OPENAI_API_KEY="sk-or-..." # After — that's the whole diff export OPENAI_BASE_URL="https://api.freerouter.com/v1" export OPENAI_API_KEY="fr_live_..."
70% to Provider A for cost, 30% to Provider B for performance or compliance. No single vendor gets all of your spend.
A provider raises prices, drops a model, or restricts usage? Flip a routing rule. Zero code changes, zero migration.
Priority ordering with automatic failover means a slow or down gateway never takes your app down with it.
Personal and team workspaces, multiple API keys, invite members, and transfer ownership when you need to.
Turn LLM procurement into a commodity routing problem. Run competitive bids, A/B test providers by real traffic, and shift volume to whoever wins — contractually free of exclusivity traps.
Talk to us"We route 10M tokens/month — who gives the best enterprise rate?" Shift volume based on who wins.
Multi-vendor routing by design removes the single-vendor relationship risk from your AI spend.
Use your own provider keys and pay providers directly — often cheaper than a gateway's markup.
A neutral layer you own — the foundation for audit logs and compliance on your terms.
Free during the preview period. Bring your own keys and start routing in minutes.
Get your key