Announcing Aria Router: three decisioners
Ten new opt-in capabilities for the OpenAI-compatible inference gateway: re-decision boundaries, trust tiers, model performance, execution evidence, soft quotas, backend failover, policy consistency and a redesigned dashboard.
- Ten new opt-in capabilities for the OpenAI-compatible inference gateway: re-decision boundaries, trust tiers, model performance, execution evidence, soft quotas, backend failover, policy consistency and a redesigned dashboard.
- Today we are announcing **Aria Router (optional enhancements)**, a release focused on evidence-led routing and production operations for our OpenAI-compatible inference gateway. It adds explicit re-decision boundaries, an afm-d three-question template, low-score protection, trust tiers, model performance, production execution evidence, API-key soft quotas, backend failover, offline policy consistency and a redesigned dashboard with Model Performance and Trust Tiers pages. All ten capabilities are **opt-in and off by default**. Read the [release news](/news/router-optional-enhancements) for endpoints and configuration, or visit the [Aria Router](/router) page.