Usage, Billing, and Logs

Use Supado logs to answer practical questions: did a request arrive, which model was requested, how many tokens were counted, and what account impact was recorded.

Before you start

• Send at least one request through https://supado.com/v1.
• Know the approximate request time and model name.
• Keep your API key identifier or application name available if you use multiple keys.
• Do not share prompt content or API keys when asking for support.

Steps

1. Find the request

Open the Supado console and filter by the time window around the request. If filters are available for your account, narrow the view by API key, model, endpoint, or status.

2. Check request metadata

Use the log entry to confirm:

• endpoint, such as /v1/chat/completions
• requested model
• HTTP status
• latency
• token usage, if returned by the upstream model
• quota or balance impact, if billing data is enabled for your account

3. Reconcile usage

For a single request, compare the log entry with the response your application received. For a batch of requests, group by API key or service so you can identify which application generated the usage.

Application deploy -> API key -> request log -> token usage -> account impact

This chain gives you a concrete path from a production change to an account-level usage change.

Investigate usage changes

When investigating an unexpected charge or quota movement, collect these fields before changing code:

• request time range
• endpoint
• model
• API key name or identifier, not the secret value
• request id, if shown
• status code
• token usage fields, if shown

Do not rely on local application logs alone. A local retry loop, queue replay, or worker restart can generate more Supado requests than a single user action suggests.

Usage can be higher than expected when retries, background jobs, queue replays, or large batches generate more requests than a single user action suggests. Token counts can also vary by model family and upstream behavior.

Supado documentation should not publish private provider cost tables or internal routing rules. If a page needs pricing or route-specific behavior, link to confirmed product material instead of guessing.