llama-dash

The admin API (/api/*) serves JSON to the dashboard. When dashboard auth is enabled, all routes below are gated except /api/auth/*. JSON GET responses support conditional ETag polling (304 on a matching If-None-Match).

Auth

/api/auth/* — Better Auth: first-user signup, username/password and passkey sign-in, session lookup, sign-out.

Models

GET /api/models — list models (merged with running state + peer info)
GET /api/models/:id — model detail (stats, events, recent requests, config snippet, key breakdown)
POST /api/models/:id/load, /api/models/:id/unload, /api/models/unload
GET /api/model-timeline — load/unload events for the timeline

Requests

GET /api/requests — cursor-paginated list
GET /api/requests/stats — req/s, tok/s, p50, error rate + sparklines
GET /api/requests/histogram — bucketed req/s histogram
GET /api/requests/:id — detail with adjacent navigation

Usage

GET /api/usage?days=1|7|30|90 — rolling 24-hour or UTC-day input/output token totals, previous-period comparison, and configured breakdowns
GET /api/usage/summary — output-token totals and series for the last 24 hours, UTC today, 7 days, and 30 days
GET/PATCH /api/settings/usage — read or update the three breakdown dimensions

See Usage for the reporting and retention behavior.

System & events

GET /api/health — upstream reachability, version, latency
GET /api/system — runtime, update status, DB, proxy, log queue, poller status, GPU details
POST /api/system/request-logs/prune, POST /api/system/database/compact
GET /api/gpu — cached GPU stats
GET /api/events — SSE dashboard invalidation/data events
GET /api/log-events — llama-swap log SSE (Logs page only)

Config & policy

GET/POST /api/config, POST /api/config/validate
GET/POST /api/keys, GET/PATCH /api/keys/:id
/api/aliases — model alias CRUD
/api/routing-rules — routing rule CRUD + reorder
/api/upstream-credentials — encrypted credential vault
/api/mcp-relays — MCP relay CRUD
GET/PATCH /api/settings/attribution, GET/PATCH /api/settings/request-limits

Playground & metrics

POST /api/playground/article-extract — fetch + Readability-parse an article URL for Speech TTS
GET /metrics — Prometheus text exporter (see Metrics)

Admin API