Models, quotas, and billing

Platform models

After provisioning, a tenant gets access to the platform model catalog plus a trial or paid quota. Models are connected by the platform and synced to the client after sign-in.

Platform model usage is metered in tokens and appears in the tenant bill.

Bring your own models

Enterprises can connect their own models:

• Self-hosted NewAPI (OpenAI-compatible).
• Other OpenAI-compatible gateways.
• Privately deployed model services.

Configure the instance URL, token, and group in the tenant console; the client syncs the model list. Usage on your instance is your responsibility; the platform handles calls and display as agreed in the contract.

Switch models

Use the model selector at the top of a session. Switching affects later messages in that session only; history is unchanged.

If a model is unavailable, the client prompts you and lets you switch to another authorized model.

Token usage

The session shows token usage for the current turn and cumulative usage; the console shows usage by user, model, and time.

Platform billing is based on NewAPI consumption logs; other statistics are informational.

Quotas

• The platform can set daily/monthly quotas per tenant (tokens or amount).
• Tenant administrators can set smaller quotas per member.
• When a quota is reached, new requests are rejected with a prompt to contact an administrator.

Billing

• Calendar-month billing, invoiced the following month.
• CSV/Excel export for reconciliation.
• Invoicing and payment follow the contract.

Common questions

No models available: confirm the role has model access, or the tenant has a model instance.

Usage jumped: check for long sessions, loops, or large files; inspect usage by user/model in the console.

Bring-your-own model unreachable: confirm HTTPS, network reachability, a valid token, and a supported version.