One endpoint, every model
POST /v1/chat/completions, OpenAI-compatible down to the error shapes. Keep your SDK, your tools and your prompts — change the base URL and the model string does the rest.
N4N AI · GATE 01
Every model. One gate.
n4n AI is a unified gateway to 240+ models behind one OpenAI-compatible endpoint. Point your SDK at api.n4n.ai, name a model per request — routing, automatic failover and per-token billing are already wired in.
Liveintelligence · api.n4n.ai · pay per token · $0 minimum
LIVE TRACE · GATE 01
Watch a request find its way
Every request resolves against a live routing table — candidate providers ranked by your directives, tried in order, rerouted on failure. This is what that looks like from the inside. The caller only ever sees one clean response.
:floorroutes to the cheapest healthy provider
:nitroroutes to the fastest
models[]falls back down your list, in order
POST /v1/chat/completionsmodelclaude-fable-5
upstream attemptsrerouted in 180 ms
anthropic · direct429 · rate limited
gcp-vertex · pool-b200 · 0.9 s
prompt 1,842 tk · completion 486 tk$0.008114debited
illustrative trace · models, providers and prices vary by route
CAPABILITIES · GATE 01
Everything between your code and 240+ models
Routing you can steer
Append :floor for the cheapest healthy route or :nitro for the fastest. Pass a models array and the gateway falls back down your list, in order, on every request.
Failover that just happens
Rate limits, timeouts and provider outages reroute to the next candidate mid-request. Your client sees one clean response; the detour only shows up in the trace.
Prepaid, per token
Load credit with PayPal or crypto — $0 minimum. Every request debits one balance at the model’s per-token rates, metered to the micro-dollar. No subscription, no seats.
Keys with guardrails
Mint as many API keys as you need, each with its own model allowlist, rate limit, expiry date and spend view. Revoke one without touching the rest.
A catalog that stays current
GET /v1/models returns 240+ models with context windows, per-token prices and supported parameters — updated as providers ship, no redeploy on your side.
ROUTE MAP · CONNECTIONS
The gateway the rest of the fleet boards through
Flow, Analytics and Desk are built to hand off to the gateway the day they board — same keys, same balance, same routing table.
n4n AI
gate 01 · live
n4n Flow
ScheduledAn AI step in any Flow automation is a single node that calls the gateway on your balance — no extra vendor, no second bill.
View the gaten4n Analytics
ScheduledToken spend, request volume and failover counts stream into Analytics dashboards, right beside revenue and pipeline.
View the gaten4n Desk
BoardingDesk drafts ticket replies through the same gateway, so support answers ride your model routing and your one balance.
View the gateA DAY AT GATE 01
A gateway’s day, minus the drama
local time · illustrative
06:12overnight embedding batch lands412 requests · one balance
09:02claude answers a customer in 0.9 s1,204 tokens · $0.026 debited
11:47provider timeout on a gpt routererouted in 180 ms · caller saw 200
16:26staging key mintedallowlist: gemini family · expires in 30 days
Is n4n AI compatible with my existing OpenAI client?
Yes. The API is OpenAI-compatible — point your SDK’s base URL at https://api.n4n.ai/v1, keep your existing code, and set the model field to any of the 240+ models in the catalog. Streaming, tool calls and the usual parameters work as you expect.
How does billing work?
n4n AI is prepaid. You load credit — there is no minimum — and every request debits one balance at the model’s per-token rates, metered to the micro-dollar. No subscription, no seats, no monthly invoice. Top up with PayPal or crypto.
What happens when a provider fails mid-request?
The gateway retries the next candidate automatically — rate limits, timeouts and outages reroute without surfacing an error to your client. You can steer the order with a models fallback array, provider preferences, or the :floor and :nitro suffixes.
Do I need the rest of the n4n fleet to use it?
No. n4n AI stands alone — one account, one API key and you are routing. As the rest of the fleet boards, the same login and balance carry across every app, but nothing about the gateway depends on them.
GATE 01 · NOW DEPARTING
Board the first departure
240+ models behind one endpoint, billed per token from one prepaid balance. No subscription, no seats, no minimum.