Every LLM call — governed, captured, costed.
Token Monitor sits in the request path as an OpenAI-compatible proxy: enforce budgets and guardrails in real time, capture the full record of every call, and re-bill customers with a markup — all from one console.
In-path gateway
An OpenAI-compatible Go proxy with virtual keys, live budgets, rate limits, fallbacks and an exact-match cache — governed at the edge, not after the fact.
Full capture ledger
Every routed call captured asynchronously — tokens, cost, latency, traces and sessions. Capture never blocks or slows the request path.
Trust by default
PII redaction before storage, per-org no-log, envelope encryption at rest, content moderation, prompt-injection guardrails, audit logs and DSAR export/delete.
Finance & re-billing
Attribute spend to customers, apply a markup, and generate invoices — sub-cent accurate, multi-currency, with CSV export.