At a glance
Route chat through a configured OpenAI-compatible upstream with a virtual key for each app. Trace reported usage by key and compare rate-table estimates when usage and a model rate are available.
Give each connected app its own virtual key and send its OpenAI-compatible chat requests through one gateway. tokenflow forwards them to the upstream you configure without forwarding the app's key. Its dashboard groups recorded requests by key, model and app label. Reported zero usage stays zero; missing usage, unavailable model rates and historical amounts are labeled separately instead of becoming made-up savings. Dollar figures use a local rate table and are estimates, not invoices. Optional subscription routing is off by default; its $0 route cost is an assumption, not a verified provider charge. Model listing and audio transcription also proxy through the gateway.