Approved models and quotas

Administrators decide which teams can use the developer API, which models it may call, and how much. Everything below lives in one place: Admin → Developer API → Management, where you pick a Team and then work through its tabs (Access, Approved models, Quotas, Keys).

Note

A developer key is scoped to a team, so all of this is set per team. A user can only create a key for a team once that team has developer access enabled (see Get an API key).

Enable it per team

1. Go to Admin → Developer API → Management.

2. Choose the team in the Team selector.

3. On the Access tab, turn Developer API access on. Members of that team can now create keys and call the gateway.

Approve models

Developers can only call models on their team’s allow-list, so the gateway can never reach a model you haven’t sanctioned.

1. On the same page, open the Approved models tab.

2. Add one or more models the team may call.

The name you approve is exactly what a developer passes as the model in a request, and what GET /api/v1/models returns for that team. If a call names a model that isn’t approved, the gateway refuses it.

Set quotas

1. Open the Quotas tab.

2. Set rate, spend, and token limits so a runaway script or an unexpected workload can’t blow the budget. Requests beyond a quota are refused rather than silently billed.

Tip

Start teams with conservative quotas and a small model allow-list, then widen as you see real usage in the cost and oversight reports.