Track and reduce OpenAI Codex spend with tiered routing and effort control.
Cheaper doesn't just show what Codex costs — it lowers it. Every request is triaged and routed to the cheapest capable model, escalating only for the hard, correctness-critical work. The gateway logs each decision so your savings are measured, not promised.
Point Codex at the Cheaper gateway (it speaks the Anthropic API), then watch the monitor:
npx cheaperapp install --all
cheaper gateway start
export ANTHROPIC_BASE_URL=http://localhost:8787
cheaper monitor