Zen

The metered gateway every Fring model request goes through.

Fring Zen is the gateway behind Fring. Signing in points the agent at it, and every model request is routed and metered there — so there are no per-provider API keys to manage, and usage for a whole workspace is visible in one place.

Zen is not an optional add-on. It is how Fring reaches a model.

Models

Zen carries a deliberately small, tested catalog rather than everything available:

minimax-m2, minimax-m3
glm-5.2, glm-5.3-flash
kimi-k2.6, kimi-k2.7-code, kimi-k3
qwen3.7-plus
deepseek-v4-pro
gpt-oss-20b

The catalog lives in the console, not in the client, so models can be added or retired without a release. fring models prints the list your install can currently reach.

See Providers for how to select one, and for what happens to direct provider credentials on a signed-in account.

Endpoints

Zen speaks the three common wire formats, so existing tooling can point at it directly:

Format	Endpoint
OpenAI chat completions	`https://fring.ai/zen/v1/chat/completions`
OpenAI responses	`https://fring.ai/zen/v1/responses`
Anthropic messages	`https://fring.ai/zen/v1/messages`
Model list	`https://fring.ai/zen/v1/models`

Authenticate with your Zen key as a bearer token. The key is issued when you sign in and is scoped to your workspace:

curl https://fring.ai/zen/v1/chat/completions \
  -H "Authorization: Bearer $FRING_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"glm-5.3-flash","messages":[{"role":"user","content":"hello"}]}'

Model ids are bare here (glm-5.3-flash), without the opencode/ provider prefix the agent uses in its own config.

Plans and usage

Fring is sold as a plan with a token allowance, not a prepaid balance:

Free — enough to run real agent tasks before deciding, with no card on file.
Pro — a monthly allowance for individual developers.
Enterprise — a shared pool across the workspace, with SSO and admin controls. Five-seat minimum.

Current prices are on fring.ai, which renders them from the live billing catalog. They are deliberately not restated here — a price list in documentation is a price list that goes stale.

To see your own plan and usage:

fring console status

For teams

Enterprise workspaces draw from one pooled allowance rather than per-user allowances, so a developer who needs more in a given week is not blocked while a colleague’s allowance goes unused. Administrators can set per-member guardrails, restrict which models members may select, and review usage per person.

See Enterprise for pooled tokens, roles, SSO and provisioning.

Data handling

For current data-handling, retention and residency terms, contact support@fring.ai. Terms depend on the upstream provider behind each model, so they are maintained alongside those contracts rather than restated here.