Teams, Budgets, and Model Whitelisting Are on Antbase
ยท5 min read
๐Ÿ›ก๏ธ

Teams, Budgets, and Model Whitelisting Are on Antbase

Antbase is now a team platform. Invite your team into a shared workspace with roles, set hard spending caps per team and per project so a runaway job can't produce a surprise bill, and whitelist exactly which models each team or project may use. All governed from one dashboard, enforced on every request.

Antbaseteamsgovernancebudgetsspend controlsmodel whitelistingRBACenterprise

Until now an Antbase workspace belonged to one person. That's fine for a solo build, but teams asked for three things before they could standardize on it: bring the whole team into one account, put a hard ceiling on spend, and control which models each group is allowed to call. All three are live today, and all three are enforced on the server โ€” not just shown in the dashboard.

Teams: one workspace, many people, real roles

A workspace is now a team. Invite people by email, they accept with one click, and they share the same keys, projects, usage, and billing. Every member has a role, and the role decides what they can do:

  • โ€ขOwner โ€” full control, including billing, budgets, whitelists, and transferring ownership.
  • โ€ขAdmin โ€” manage members, API keys, budgets, and model whitelists. The day-to-day operator role.
  • โ€ขMember โ€” use the workspace and its keys; no governance controls.
  • โ€ขViewer โ€” read-only access to usage and logs.

Your existing account became the owner of your workspace automatically โ€” nothing to migrate. Add teammates from the new Team page; they show up the moment they accept.

Budgets: a hard stop, not a warning email

The point of a budget is that it holds. Set a daily and/or monthly cap on a team โ€” or on an individual project โ€” and once spend for that window reaches the cap, Antbase stops accepting requests and returns a clear error until the window resets or you raise the cap. No overage, no surprise invoice. The cap is checked on every request, before any model is called.

  • โ€ขCaps apply per team and per project, so one noisy project can be contained without throttling the rest.
  • โ€ขWindows are calendar day and calendar month (UTC); spend resets automatically when the window rolls over.
  • โ€ขA blocked request returns HTTP 402 with a message telling you which cap was hit and when it resets.
json
{
  "error": {
    "type": "insufficient_quota",
    "code": "BUDGET_EXCEEDED",
    "message": "Team daily budget reached: $50.00 of $50.00. Requests are paused until the next UTC day or the cap is raised."
  }
}

Model whitelisting: control what each team can call

Set an allowlist of models on a team, on a project, or both โ€” and Antbase will only ever route to models on that list. Pin a model that isn't allowed and the request is rejected with a clear MODEL_NOT_ALLOWED error; send a virtual model like ant:auto and the router simply picks the best allowed model. A project's allowlist is intersected with the team's, so a project can only ever narrow what the team permits, never widen it.

This is how you keep an experimental team on cheap open models, hold a regulated project to a specific approved set, or make sure nobody accidentally routes production traffic to a model you have not signed off on.

bash
# Team allowlist excludes gpt-5; pinning it is rejected up front.
curl https://antbase.ai/v1/chat/completions \
  -H "Authorization: Bearer $ANT_API_KEY" \
  -d '{"model":"gpt-5","messages":[{"role":"user","content":"hi"}]}'
# โ†’ 403 MODEL_NOT_ALLOWED

# ant:auto still works โ€” it routes only within the allowed set.
curl https://antbase.ai/v1/chat/completions \
  -H "Authorization: Bearer $ANT_API_KEY" \
  -d '{"model":"ant:auto","messages":[{"role":"user","content":"hi"}]}'

One place to govern it

Members and roles, team and project budgets, and model allowlists all live in your dashboard. Set them once; they apply to every key and every request the team makes.

Try ANT routing today

Drop-in OpenAI-compatible API. Change your base URL and every request gets intelligent routing across 30+ providers.