Features
From unified proxy to org governance, cost control, and compliance—core needs for enterprise LLM gateways.
Unified proxy & routing
- OpenAI-compatible: /v1/models, chat/completions (streaming), embeddings
- Anthropic-compatible: /v1/messages, count_tokens
- Model catalog + channels + mapping + failover; sticky key pools & cooldown
- Vision pass-through / text conversion + cache; end-to-end X-Request-Id
Identity & API key governance
- Local accounts with forced change of weak default passwords
- OIDC SSO (optional password login; real IdP ready)
- Multi-key per user: notes, expiry, revoke, revoke-all
- Keys bindable to projects and IP allowlists
Org, departments & projects
- User quotas & model grants; department packs (optional apply)
- Department admins: members + read-only billing
- Project workspace: members, intersection ACL, X-Project-Id / key bind
Metering, pricing & budgets
- Price-table estimates; filter billing by user/dept/project
- Split retention for hot/billing/audit; pricing versions & monthly snapshot lock
- Site/dept/project monthly budgets; 80% webhook alerts + optional hard limit
Compliance & security
- Body policy: off / summary / full with secret redaction
- Confirm-to-view request bodies with audit
- Upstream channel keys encrypted at rest (enc:v1)
- Admin IP allowlist; login success/failure audit
Observability & ops
- Overview, trends, request detail, speed, vision stats
- Prometheus: GET /metrics
- Scripted deploy, health checks, channel probes