2 models, one API key for all of them. Compare by price, capability and measured performance.
Requests reach each model’s official API, and a third-party signed receipt lets you verify which model answered.
You’re billed for the usage the provider reports, and 2 models cost less than official prices. Every request has an itemized receipt.
If a provider becomes unavailable, requests fail over to another; provider uptime is published.
Native OpenAI, Anthropic and Gemini APIs; Claude Code and Codex connect by changing the base URL.