Point any tool's base URL at the Cheaper gateway — it speaks both the Anthropic and the OpenAI-compatible API, so it routes far beyond Claude. See how each one connects →
Every request is classified before it runs. Simple work is answered at the cheapest tier, and the model you asked for is a ceiling — Cheaper only ever matches it or goes cheaper.
Concurrency, security, proofs, high-stakes judgment — these are a downgrade veto, not an upgrade trigger: Cheaper never cuts them to a cheap model to save money, and by default it never spends more than the tier you asked for.
The gateway records the model you asked for and the model it served, so its figures are measured. Figures read from your local chat history are estimates, and are labelled "about".