Know what every feature, customer and API key costs. LLM Gateway prices each call, attributes it and stops spend before it reaches the ceiling you set.
Token spend is hard to see because it is spread across providers, keys and features, and it is discovered at the end of the month in an invoice nobody can decompose.
LLM Gateway prices every request at the moment it happens, against the current catalog price for the model that served it, and stores the result with the key, project and model that produced it.
Each call leaves a record that can be aggregated any way you need.
Alerts warn when a project approaches its budget, and hard limits reject further traffic instead of quietly spending past the ceiling. Limits can be set per key, project or environment.
Once cost is measurable it can drive routing: send high-volume, low-risk traffic to cheaper models, keep premium models for the paths that justify them, and let price ceilings move traffic automatically when a vendor changes its list price.
Short answers about the gateway, the providers and how billing works.
Costs are computed from the token counts reported by the provider and the catalog price in effect for that model at the time of the request.
Yes. A request can carry a customer or feature tag, and reports can be grouped by that tag, by key, by project or by model.
The key or project stops being served until the limit is raised or the next period starts, and the event is recorded in the usage log.
Usage and cost records can be exported for finance reporting or pushed into your own warehouse from the Scale plan.
Sign in with Google, open the playground and price a request against the whole catalog before you change a line of production code.
