Getting Started
Fluxtab reads billing and usage from OpenAI, Anthropic, and OpenRouter, then shows spend in one dashboard with Slack (and other) alerts when you cross a threshold.
It does not sit in front of your API traffic. Production requests keep going directly to your providers. Connected keys are read-only by design — Fluxtab can alert you, never block a request.
Why use it
You want one place to see MTD spend across providers, a simple forecast of where the cycle is heading, and a warning before you blow a budget — without putting Fluxtab in the request path.
Setup
- Create a Fluxtab account and open Settings → Connections.
- Add a provider (OpenAI, Anthropic, or OpenRouter) with an admin / org key. For OpenRouter, paste a Management key from OpenRouter → Settings → Management Keys — a normal inference API key will not work for usage history.
- Open Settings → Alerts and set one monthly USD threshold. Fluxtab warns at 50%, 80%, 100%, and 120% of that number.
- Add at least one alert channel so those warnings have somewhere to go.
- Optionally connect Slack or Discord bots so you can run
/fluxtab spendfrom chat. - Open the Dashboard and run Sync now (or wait for the scheduled sync) to pull usage.
Free plans sync about every 6 hours; Pro syncs about every hour. This is not live, same-minute monitoring — spikes show up same-day (faster on Pro).
Budgets
The monthly USD threshold under Settings → Alerts is the budget Fluxtab compares against. Threshold alerts fire at 50 / 80 / 100 / 120% of that number. The AI spend budget gate uses the same threshold (or a per-request budget= override) as a budget trajectory check when month-to-date spend is already over.
Fluxtab cannot hard-stop provider traffic. Treat this budget as early warning — pair it with alert channels so someone actually sees it.
What Fluxtab will not do
Treat Fluxtab and your provider’s own budget settings as early warning, not a guarantee. As of 2026, OpenAI’s account budget setting notifies but does not hard-stop requests, and Anthropic has no equivalent cap for standard API keys. Set provider-native limits yourself if you need spend to stop.