QVeris Model Gateway

Version: 2026-09-04

QVeris Model Gateway exposes an OpenAI-compatible model API. Use your QVeris API key for authentication and settle successful calls against your Credits.

Base URL

Model endpoints start with /v1, for example https://aigateway.qveris.ai/v1/chat/completions.

Prerequisites

1. Register a QVeris account and create an API key (starts with sk-) in Account → API Keys.
2. Keep a positive Credits balance. Successful calls are charged by actual usage; failed calls are not charged.

List available models

export QVERIS_API_KEY='sk-your-key'

curl -sS https://aigateway.qveris.ai/v1/models \
  -H "Authorization: Bearer $QVERIS_API_KEY"

The data[].id values returned by GET /v1/models are the public model IDs to use in calls. Always trust the live model list instead of hard-coding upstream provider model names.

Chat completions

curl -sS https://aigateway.qveris.ai/v1/chat/completions \
  -H "Authorization: Bearer $QVERIS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [
      {"role": "user", "content": "Introduce yourself in one sentence."}
    ],
    "max_tokens": 64
  }'

The response follows the standard OpenAI Chat Completions structure and also includes qveris_billing settlement information (see Billing).

For streaming, send "stream": true. After the request finishes, the gateway appends an event: qveris.billing event immediately before data: [DONE].

Other supported model APIs

Only call an endpoint with a model whose capabilities in GET /v1/models support it.

Authentication and request headers

Every /v1/* request must include:

Authorization: Bearer <QVeris API key>

Save the X-QVeris-Call-ID response header from every call. It is the cross-reference ID for audit, billing, and support.

Billing

A successful non-streaming response includes qveris_billing:

{
  "qveris_billing": {
    "call_id": "d2ea8e43-dbb1-4846-ba1c-2acac1abadad",
    "credits_charged": 0.025,
    "cost_usd": 0.00005082,
    "credits_per_usd": 500,
    "usage_estimated": false,
    "pricing": {
      "mode": "expression",
      "prompt_usd_per_million_tokens": 0.22,
      "completion_usd_per_million_tokens": 0.66
    }
  }
}

Key points:

• The current exchange rate is 1 USD = 500 Credits.
• Admission checks only that the balance is positive; success settles actual usage and clear failures are not charged.
• credits_charged is the actual Credits deducted and may differ slightly from the theoretical value because of minimum precision.
• usage_estimated=true means the upstream usage was incomplete and the gateway settled with a conservative estimate.
• The QVeris usage history is the authoritative billing record.

Errors

Gateway errors use a uniform envelope:

{
  "error": {
    "message": "the requested model is not available",
    "type": "gateway_error",
    "code": "model_not_found"
  }
}

FAQ

• Which model ID should I use? Use the id returned by GET /v1/models.
• Why do I get 404? Confirm the model ID is correct and enabled. The model catalog refreshes every 30 seconds.
• Why do I get 429? Respect the Retry-After response header and lower concurrency.
• How do I get support? Keep the X-QVeris-Call-ID response header and share it with the support team.
• How should I protect my key? Do not paste full keys into docs, logs, or tickets. Inject keys through environment variables.