Platform AI

Vibeforce no longer has this page. It holds no AI provider key at all.

The platform's own keys, the models it runs on and the catalog it offers organizations are one set of settings for the whole ecosystem, kept once at the Vibeforce account service. Vibeforce had its own copies; keeping them would have meant two tables, two screens and two ways for a key to get out.

The admin sidebar carries two links out instead:

Link	Where it goes	What it holds
Platform AI ↗	the account service's /admin/ai	The keys and models the platform itself runs on. The platform's bill.
Model catalog ↗	the account service's /admin/model-catalog	The models an organization can be managed by, offered for reference.

Both open in a new tab, and neither ever shows as the active page — they are doors, not sections. Getting in needs a platform administrator account at the account service; being a Vibeforce platform admin is not by itself enough, and the two are worth keeping straight.

A third door sits beside them: AI tokens ↗, at the account service's /admin/tokens — the token allowance every account is metered against, across all three products. It replaced the Credits page. See The token allowance, for administrators.

The admin area that is left is four pages: Overview, Skills, Designs and Settings.

What the platform's keys are for now

This changed with the same release. An installation's key is no longer what every build runs on — it is what a build runs on when that build's organization has chosen managed mode. See AI configuration.

The organization's mode	Which key runs the build	Metered?
Managed by us	the platform's, configured at the account service	Yes — against the organization owner's token allowance, and refused once it is spent
Self hosted	the organization's own	No — recorded, never limited

So the platform key is still what makes the installation work out of the box, and still what most builds run on. It is no longer the only way anything can build: an organization with its own key needs nothing from this page.

If no platform model is configured, managed mode cannot be chosen. Managed by us is greyed out on every organization's AI page — "Nothing is on offer yet — the platform has no model configured" — and on an installation in that state, nothing builds until an organization brings its own key. Configuring a platform key remains the first thing to do on a fresh installation.

Provider keys, at the account service

Checked against the provider before they are stored, then encrypted. Only the last four characters are ever shown. Adding a key for a provider replaces the one it already has.

Six providers can hold a key: OpenAI, Anthropic, DeepSeek, xAI (Grok), Google (Gemini) and Azure OpenAI — plus an OpenAI-compatible server of your own, addressed by base URL. That is two more than Vibeforce's own page ever offered.

A key is verified live against the provider before it is stored, so a rejected key never reaches storage. The same two cautions apply as before:

• One key per provider. Adding a second replaces the first.
• Removing a key removes every model bound to it, including ones added by hand. With one key per provider there is nothing left to run them on.

Unticking active is the non-destructive way to take a provider out of service.

Keys are encrypted and there is no plaintext fallback: with no encryption key configured, the store refuses to save and says so on every read.

Models

Browse models asks the provider for its live catalog, minus what is already there; capabilities are worked out from the identifier, because providers report names and not capabilities, and anything unfamiliar is added as a plain chat model rather than hidden. There is also an add-by-identifier path.

The models table renames, activates and stars.

• The starred default is what a managed build runs on, and it must support tools, or the builder cannot read or write a file. One default per kind; starring clears every other star of that kind and activates the new one.
• There is no pricing to set. A model record still carries a credit_cost from the retired credit system, but nothing charges against it — a managed build is counted in raw tokens against the account's allowance. See What a build costs.

The model catalog

The second link. Reference data: what the platform offers organizations. It holds no key and can call nothing.

An organization in managed mode sees its contents on its own AI page under The platform catalog, so it can see what it is getting. It picks nothing from it — what runs is the platform's starred default.

Entries carry a provider, a name, an identifier, a kind and optional notes, and are either offered or retired. An entry's identifier cannot be edited: retire it and add the new one instead.

Messages users see when this is wrong

Message	What to fix
No models are configured yet.	The catalogue is empty.
Models exist but none are switched on…	Tick Active on a chat model.
The active models cannot call tools, so the builder cannot read or write a file.	Activate one with the Tools badge.
No active <provider> API key is configured…	The starred model's provider has no active key.

Each of these can now come from either scope. If it names an organization's own settings, the fix is on that organization's AI page and not here.

Next

• What an organization sets for itself.
• The token allowance, for administrators.
• Skills.