The free tier

Every signed-in account can use Hylon for free, up to a daily token allowance (currently 100,000 tokens per day, roughly a few hundred short chat turns). The allowance:

• Applies to input plus output tokens combined.
• Resets every day at midnight UTC.
• Requires no subscription and no credit card.

The allowance is operator-tunable, so treat the number in your GET /api/v1/hylon/me response (free_daily_tokens) as the source of truth.

Paid tiers

Paid tiers add a monthly included token quota, higher effort ceilings, and more concurrent sessions. The current catalog:

Subscribe from app.hylon.ai (Settings, Billing) which runs a Stripe checkout, or start one programmatically with POST /api/v1/hylon/checkout (plan_id from the plans endpoint, plus optional success_url, cancel_url, email). If Stripe is not configured on a deployment, checkout answers 503 rather than pretending to work.

Credits and per-token pricing

Beyond the free allowance and any included quota, usage is charged against a prepaid Hylon credit balance. Credits are a separate pool used only by Hylon AI: they never mix with VPN wallet balances, ORB tokens, or scan credits.

Rates are per model, in credits per million tokens. Current pricing:

The live rates are public at GET https://api.orbai.world/api/v1/hylon/pricing. A model with no active pricing is never served silently for free; such requests are rejected.

How metering works

Every chat completion is metered by the serving gateway and settled against your account in a fixed order:

Each request is recorded exactly once (settlement is idempotent per request id). Review your history any time:

• In the app: app.hylon.ai, Settings, Billing shows your tier, credit balance, and a monthly usage bar.
• Via the API: GET /api/v1/hylon/usage returns your balance and recent usage events (model, tokens, cost in credits, charge source), and GET /api/v1/hylon/me returns your resolved entitlement.

API keys

API keys give scripts and integrations access to Hylon with your entitlement, without sharing your login.

• Format: hyl_sk_ followed by an 8-character key id and the secret body, for example hyl_sk_1a2b3c4d_....
• The full secret is shown exactly once, at creation. Only a hash is stored; if you lose the key, revoke it and create a new one.
• Listing your keys shows the non-secret prefix only.
• Keys can be named, can carry an optional expiry (expires_in_days), and can be revoked at any time.
• A key cannot mint other API keys and cannot mint node setup codes; both require a signed-in session.

Manage keys in app.hylon.ai (Settings, API Keys) or via the API:

# Create (session auth; the secret appears once in the response)
curl -X POST https://api.orbai.world/api/v1/hylon/api-keys \
  -H "Authorization: Bearer <YOUR_SESSION_JWT>" \
  -H "Content-Type: application/json" \
  -d '{"name": "my-script", "expires_in_days": 90}'

# List (prefixes only)
curl https://api.orbai.world/api/v1/hylon/api-keys \
  -H "Authorization: Bearer <YOUR_SESSION_JWT>"

# Revoke
curl -X DELETE https://api.orbai.world/api/v1/hylon/api-keys/123 \
  -H "Authorization: Bearer <YOUR_SESSION_JWT>"

Calling the AI gateway

Programmatic chat goes through the Hylon serving gateway, which is OpenAI-compatible:

• Base URL: https://hylon-gateway.ambitiousmushroom-c83407ea.eastus.azurecontainerapps.io
• POST /v1/chat/completions (streaming and non-streaming)
• GET /v1/models lists the served model

Authenticate with either your session JWT or a Hylon API key in the Authorization header:

curl https://hylon-gateway.ambitiousmushroom-c83407ea.eastus.azurecontainerapps.io/v1/chat/completions \
  -H "Authorization: Bearer hyl_sk_..." \
  -H "Content-Type: application/json" \
  -d '{
        "model": "Qwen/Qwen2.5-3B-Instruct",
        "messages": [{"role": "user", "content": "Hello, Hylon"}]
      }'

The gateway checks your entitlement before serving (401 without valid credentials, 402 when nothing can cover the request) and reports the exact token counts for settlement after the response completes. Streaming responses are metered too.