Get started
Follow the steps for the capability you are integrating.
Connection details
Choose an endpoint
Native endpoint Recommended
Use the model provider's native format, such as the Anthropic API for Claude. This preserves the complete feature set without format conversion.
OpenAI-compatible endpoint
Use any supported model through an OpenAI-compatible interface. For existing OpenAI integrations, update the base URL.
Compatibility considerations
The compatibility endpoint converts between formats. Some model-specific capabilities, such as the full extended-thinking structure, may be simplified.
Anthropic native endpoint (recommended)
Python(anthropic SDK)
OpenAI-compatible endpoint
Python(openai SDK)
Usage
Review your account balance and API usage.
API usage
Usage by model
Billing reconciliation
Export itemized billing records by request ID, timestamp, model, streaming mode, token usage, and cost. The CSV is compatible with Excel and includes both streaming and non-streaming requests from the metering ledger.
Models
View the models and prices available to your account. Availability is synchronized with the gateway, and cache pricing reflects the provider's published rates.
How these prices are applied
Text models
Billed per token, with input and output counted separately. Prompt caching, where the model supports it, is billed at the separate rates shown on the card. Prices are the model provider's published rates.
Logs
Per-request records for reconciliation: model, timing, token counts and cost for every API call.
API Keys
Manage API keys for separate projects or integrations. Each key can have its own spending limit, while the account balance remains the overall limit.
Account
Balance alerts
Loading current usage…