Dashboard Workflows

The Dashboard is where users create keys, control spend, test prompts, and troubleshoot requests.

Create an API key

Open Dashboard -> API Keys.

Fields:

After creating a key, copy the full value immediately. The key can also be revealed later from the key table, but users should still store it in a secret manager or environment variable.

Choose model scope

Use the least restrictive setting that still controls risk:

• All models: best for early exploration because future public models stay usable.
• Chat models: good for chat-only apps.
• One model ID: good for production services that should not drift to other models.

If a scoped key cannot access a model, the public API returns model_not_found.

Reveal, copy, revoke, and enable

API Keys supports:

• Reveal full key.
• Copy full key.
• Revoke an active key.
• Enable a disabled key.

If a key leaks, revoke it, create a replacement, update environment variables, and check Usage Logs for suspicious activity.

Use Playground

Playground lets a signed-in user test a prompt without copying the key into local code.

It uses:

• one active API key from the account;
• one publicly available chat model;
• system message;
• user prompt;
• temperature;
• max_tokens.

After a successful run, Playground shows the response and equivalent curl, Python, and JavaScript snippets.

Use Usage Logs

Usage Logs are the first place to look after a request fails.

Available filters include:

• request ID;
• model;
• API key;
• success or error status;
• date range.

The table and detail view show:

• timestamp;
• key name;
• model ID;
• prompt tokens;
• completion tokens;
• cost in USD and quota;
• latency;
• request ID;
• error content when available.

CSV export is available for filtered logs.

Use Billing

Billing shows:

• current balance;
• remaining quota;
• recent spend;
• top-up packages;
• custom top-up amount;
• payment history.

If payment is unavailable, pending, or disputed, contact support instead of retrying the same checkout repeatedly.