Concepts

• OpenAI-Compatible API

  The platform standardizes on the OpenAI Chat Completions / Responses format and translates requests to whatever each upstream provider expects (Anthropic, Gemini, DeepSeek, xAI,…

• Routing & Failover

  Each model may be served by multiple upstream channels (pools). Requests are routed as follows: Cost-first: automatically selects the cheapest available channel Automatic failover:…

• Billing Model

  Per token: input price × input tokens + output price × output tokens Per call: image / video models have a fixed per-request price…

• Rate Limits

  Each model has independent RPM / TPM limits. Exceeding them returns 429. Check the current limits on each model’s detail page For batch workloads,…

• Data & Privacy

  Zero retention by default: request/response content is never stored, cached, or used for training Enterprise audit logs: SOC 2-ready audit logs on request TLS…