Give every team a budget.Never get a surprise bill.
Hard caps per key, user, team and organization. Alerts before a limit hits. Every dollar attributed to the model, feature and person that spent it.
Caps that enforce themselves
Set a limit on a key, a user, a team or the whole organization. At the cap, the gateway stops spending. No cleanup script, no month-end surprise.
Hear about it before finance does
Threshold alerts by email or webhook. A runaway agent becomes a notification at 80%, not a discovery at the invoice.
- Thresholds you choose50%, 80%, 95%, not only the cap
- Per budget and per keyLoose on a sandbox, tight on production
{
"event": "budget.threshold_reached",
"scope": "api_key",
"name": "sk-data-team",
"threshold": 95,
"spend": 1905.00, "limit": 2000.00
}Every dollar has a name on it
Cost by model, team, user and key out of the box. Tag requests with your own metadata and the same chart splits by feature, customer or experiment.
Cut the cost of every request
Turn on automatic prompt caching with one flag and the gateway places the cache breakpoints for you. Moving a route to a cheaper model is a string change. Both savings land in the same dashboard as the spend.
response = client.chat.completions.create( model="anthropic/claude-opus-5", messages=messages, extra_body={"requesty": {"auto_cache": True}}, ) # repeated context is billed at the cached rate
One number, and it sits on top of the model cost
No per-seat pricing, no minimum spend. You pay for what your application spends, plus 5%.
Every feature included: routing, caching, failover, guardrails, analytics and audit logs. No seat fees, no feature tiers.
See pricingKeep the pricing you have negotiated with each provider and still get routing, limits and observability across all of them.
How BYOK worksSSO, SCIM, private regions, a signed DPA and a named contact. Procurement paperwork handled by people, not a form.
Speak to a founderPut a ceiling on every key
Budgets, alerts and attribution on every request through the gateway.