For private endpoints, there are no limits or rate limits. Resources are allocated exclusively, billing is based on the actual server runtime, so artificial restrictions on request volume or frequency are not applied.
For public endpoints, limits apply at the user account level and the associated API key. Under free access, registering additional accounts or generating new keys to bypass established quotas is prohibited. Limits do not apply to individual models or specific requests — all quotas are aggregated per user within public access.