Skip to main content

Billing, Usage, and Logs

Replace {baseUrl} with your platform API root, without a trailing / or /v1. Prices and quota conversions are deployment-specific; the current console is authoritative.

Billing​

The platform deducts pay-as-you-go charges from your account balance. The source guide describes multiplier-based billing:

Quota used = input tokens × model multiplier
           + output tokens × model multiplier × completion multiplier

The pricing page lists model_ratio and completion_ratio. A completion multiplier of 5 means an output token costs five times as much as an input token under this formula.

The source deployment reported quota_per_unit = 500000, equivalent to 500,000 quota units per CNY 1. Check your console for the current conversion and any model-specific billing rules.

View usage in the console​

Open the console to view:

  • Request logs: filter by time, model, and token to inspect token counts and individual charges.
  • Dashboard: view usage trends and model distribution.
  • Token usage: inspect each token's cumulative usage in the token list.

Usage APIs​

Use a management access token, not a model API key:

# List tokens and inspect their usage
curl "{baseUrl}/api/token/?p=0&size=10" \
  -H "Authorization: Bearer <access-token>"

# Retrieve request logs
curl "{baseUrl}/api/log/self?p=0&page_size=20" \
  -H "Authorization: Bearer <access-token>"

# Retrieve account statistics
curl "{baseUrl}/api/data/self" \
  -H "Authorization: Bearer <access-token>"

Replace <access-token> with the management credential generated after signing in.

Control costs​

  1. Create separate tokens for tools and give each a quota cap.
  2. Use inexpensive models for translation and completion, with a token-level model allowlist.
  3. Set caps for coding tools such as Cline and Cursor, which can consume large amounts of context.
  4. Review logs regularly for unusual request frequency.