n4nAI

Every token,
read on the meter.

n4n AI is a single OpenAI-compatible endpoint in front of 240+ models, billed to the micro-dollar. Prepaid credits, no subscription required, no minimums — you only pay for what actually lands.

one endpoint · pay per token · prepaid credits

Reliability

When a provider fails, you don't.

Every request tries the cheapest healthy provider first. If it rate-limits or goes down, n4n reroutes to the next one automatically — same request, same response shape, no code change on your side.

Zero cost on reroutes

Rate limits and outages retry on the next provider automatically. You're only billed for the request that actually lands.

Every request, every time

Fallback runs on 100% of requests, not just ones you opt into — reliability isn't a paid tier.

For open-weights models

Pay list price. Get most of it back.

Open-weights models carry a load factor discount of up to 90% off. Requests bill at full list price when they run; once the hour closes, n4n credits the discount back to your project. How deep it goes is set by your plan's floor.

One open-weights model · discount factor across a day
0%25%50%75%100%Pro floor · 75%00:0012:0024:00
  • Model factor, set by an n4n operator
  • What you receive — min(factor, your plan floor)

Illustrative shape only — not live data, and not a fleet readout. n4n serves open-weights models through upstream providers; the factor is set per model by an operator and informed by provider supply signals. Peak shown 90% off, clamped to 75% on Pro.

See how load factor works →

For high-volume workloads

Same call. A little patience. A better rate.

Instant batch inference isn't a separate endpoint — it's one parameter on the normal chat completions call. Add ibi and n4n holds your request in a shared queue with other opted-in traffic for that model, dispatching it upstream as a batch once the queue's ready. You still get back one response through the same call — just a little later, and priced lower.

Instant batch inference — live queue

4Requests queued right now
~280 msAdded latency this cycle

Illustrative — a simulated queue, not live telemetry. queue_factor trades latency for a deeper discount.

See how instant batch inference works →

One endpoint

One key, every provider.

Point your existing OpenAI-format SDK at api.n4n.ai/v1 with an n4n_ key and reach Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek and more — no separate SDKs, no separate keys to manage.

Drop-in compatible

If it speaks the OpenAI chat completions format, it already works — swap the base URL and key, nothing else changes.

240+ models

Browse live pricing and context windows before you commit a key.

Browse the catalog →

Why n4n AI

Metered like current, not estimated like a quote.

Usage-based billing

Every token is metered in micro-dollars against one prepaid balance. Prepaid credits never expire, and no plan is required.

Routing and fallback

Requests land on the cheapest healthy provider first. Rate limits and outages reroute automatically — failed attempts are never billed.

One key, every provider

Keep your OpenAI SDK. Point it at api.n4n.ai/v1 with an n4n_ key and reach Anthropic, OpenAI, Google, Meta, Mistral and more.

240+ models, one catalog

Browse live pricing and context windows before you commit a key.

How billing works

Top up. Send requests. Read the meter.

01

Load credits

Top up prepaid credits with PayPal or crypto, from $1. Credits never expire.

02

Send requests

Point your existing OpenAI-format client at api.n4n.ai/v1 with an n4n_ key. Streaming included.

03

Read the meter

Every response is billed as prompt + completion tokens at the model's listed price, accounted in micro-dollars.

curl https://api.n4n.ai/v1/chat/completions \
  -H "Authorization: Bearer n4n_your_key" \
  -H "Content-Type: application/json" \
  -d '{"model":"anthropic/claude-sonnet-5","messages":[{"role":"user","content":"Hello"}]}'

Example

One request, read off the meters.

Served by

anthropic (priority 0)

200 OK · 812 ms

Standby

openai (priority 1)

openrouter (priority 2)

Tokens

prompt 1,204 tok · completion 356 tok

Meter

$0.008442

Cost

$24.98

Balance

Illustrative — figures from a single real request routed through n4n AI, shown exactly as billed.

Pricing

Credits in, tokens out. Nothing else.

01 · top up

Load credits from $1

Pay with PayPal or crypto. Prepaid credits never expire and there is no monthly fee.

02 · send requests

Every token is metered

Each response is billed as prompt + completion tokens at the model's listed price, in micro-dollars.

03 · fallback included

Pay only for what lands

Rerouted attempts on rate limits or outages cost nothing.

04 · or add a plan

Four plans, per project

From $10/mo. Each grants credit equal to its price and unlocks a deeper load factor discount on open-weights models.

Compare the plans →
See the full model catalog and prices →

Start building on n4n AI.

Sign in, create a key and send your first request in under a minute.