acumen: model gateway status

import StatusPill from ‘../../../components/StatusPill.astro’;

This page summarizes acumen behavior described in repository docs. It is not a current availability check.

Snapshot (from prior docs)

• acumen is described as a Go gateway with a Postgres store, served behind ai.tcp.systems (public origin verified 2026-10-21).
• It routes aliases to cost classes and seats, meters usage, and applies x-acumen-max-class ceilings.
• It exposes OpenAI and Anthropic-shaped endpoints on the same listener.
• The X-Acumen-Served-Model header names the vendor model that answered (code date 2026-11-04).

Before relying on acumen, verify the current host, deployment, and endpoint behavior and record the date and result.

The gateway is one row on the status page, probed at GET /readyz: an empty 200 there reads as down, not healthy. Read the state vocabulary before trusting a green row.

For a complete migration guide covering base URLs, key handling, model aliases, streaming, error handling, and failover, see Migrating a client to acumen.

Next: send a request, read the API reference, or use the acumen runbook for operator checks.