AAETHEL.AI
CHECKING APIAPI quickstart ↗
AETHEL V1 · PRIVATE API PILOT

One endpoint
for AI inference.

An OpenAI-compatible gateway with customer API keys, model controls, per-key daily quotas, and durable usage records. Requests go to the inference provider configured for this deployment.

Read the API quickstart ↓Check service health ↗
i The configured provider’s rate limits and availability still apply.
SERVICE STATUS

Inference gateway

CHECKING
A
Checking service configuration…

Contacting the gateway health endpoint.

USAGE STORECHECKINGDurable request records
INFERENCE PROVIDERCHECKINGGroq backup: CHECKING
DEFAULT MODELgpt-4o-miniOperator-configured allowlist
AVAILABLE API ROUTESV1
GET  /api/healthPOST /api/v1/chat/completionsPOST /api/v1/keys ADMIN

Live status loads from this deployment. Run the quickstart to configure provider access and issue customer keys.

GATEWAY FEATURESAPI KEY AUTHMODEL ALLOWLISTSDAILY QUOTASTOKEN USAGE RECORDS
A BETTER DEFAULT

AI should be ready
when your users are.

Give applications one authenticated interface for configured model requests and durable usage tracking. Provider rate limits and availability still apply.

× THE BOTTLENECK

Centralized AI
gets in the way.

  • 01 Message caps and rate limits at peak hours
  • 02 One provider outage can stall your product
  • 03 Limited visibility into request-level token usage
→
✳ THE AETHEL APPROACH

Build with
clear controls.

  • 01 Issue and disable customer API keys
  • 02 Set a daily request limit per key
  • 03 Record provider-reported token usage
REQUEST PATH · V1

One gateway.
Optional backup routing.

The gateway validates each API key, enforces a daily request quota, and records usage. When enabled, mapped requests can fall back from an OpenAI-compatible primary to Groq after a primary rate-limit response. This reduces reliance on one provider but cannot eliminate limits or outages; distributed GPU routing is not yet available.

Explore the architecture ↗
REQUEST PATH 429 FAILOVER · OPT-IN
⌘Your appOne API endpoint
✳Aethel gatewayAuth · quota · usage
CONFIGURED PROVIDER UPSTREAM
AUTHENTICATED·METERED·PROVIDER LIMITS APPLY
THE FIRST WORKING RELEASE

Useful controls.
Visible boundaries.

Start with an operational API path, then add providers, billing, and distributed compute as those systems are built.

01
⌁

Customer API keys

Create named keys through an admin-only endpoint. Store only key hashes and reveal each key once.

SHA-256 HASH STORAGE
02
◉

Usage metering

Record request outcomes and provider-reported token counts in a persistent database.

SUPABASE POSTGRES
03
↗

Per-key daily quota

Enforce an atomic daily request limit before forwarding traffic to the configured provider.

CONFIGURABLE REQUEST LIMIT
FOR DEVELOPERS

Make your first request.

V1 is a private API pilot. Customer billing is not available yet; inference uses provider capacity billed to Aethel. Access keys are issued by an administrator. Never expose provider or admin secrets in browser code.

Open the API quickstart ↗
CHAT-COMPLETION.SHOPENAI-COMPATIBLE
# Send an authenticated request
curl https://www.aethelai.online/api/v1/chat/completions \
  -H "Authorization: Bearer $AETHEL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-4o-mini","messages":[{"role":"user","content":"Hello"}]}'
Requires deployment configuration · quotas and provider caps apply