Dashboard Tour

Every dashboard page in the order a team uses them, from first API call to a governed multi-seat account.

The dashboard at https://anyrouter.dev/dashboard has twelve pages. Do the Day-1 path first; the rest can wait for real traffic.

Day 1: first call

  1. Settings (Account settings). Check email and name, enable two-factor authentication before creating keys, set your timezone and default workspace.
  2. Keys (API keys). Click New key, name it (for example laptop-dev), and copy the sk-ar-... value into .env; it is shown once. Leave limits blank for now. The last-four suffix identifies the key in logs.
  3. Quickstart (Quickstart). Run the curl command against https://anyrouter.dev/api/v1/chat/completions with Authorization: Bearer sk-ar-.... A 200 with choices[0].message.content means it works. On errors, keep the requestId and cfRay from the body for support.
  4. Logs (Request logs). Within seconds a row shows model, tokens, cost_credits and the key suffix. Expand it for the request, response (subject to Privacy settings), latency and serving upstream.

Week 1: spend and defaults

  1. Usage Explorer (Usage Explorer). Set the range to 7 days and group by model, key, app tag or status to see what drives spend. Export CSV to share.
  2. Credits (Credits and billing). Two cards: plan credit (monthly, from your plan: Go $4, Pro $20, Pro+ $100, Max $300, plus referral or promo bonuses; Free has none) and bought credit (purchased, never expires). Usage draws plan credit first. Top up manually and enable low-balance email alerts with 3+ days of runway; automatic top-up is not available yet.
  3. Presets (Presets). Save a model, system prompt, sampling parameters and tools as a Preset and call it as @preset/my-team-summarizer. Changing the preset changes every caller.
  4. Privacy (Privacy controls). Choose full payloads, metadata only, or nothing, and a retention window. Tighten these first if you process regulated data. Zero-retention modes make debugging harder.

Scaling: teams and governance

  1. Organizations (Organizations). A shared workspace with its own credits, keys, presets and logs. Invite teammates by email with roles owner, admin or member; only owner can change billing or top up. Each teammate mints their own key; never share a personal key.
  2. BYOK (BYOK). Attach your own upstream provider keys (OpenAI, Anthropic, xAI, Z-AI and others), encrypted at rest. Traffic still flows through AnyRouter but the provider bills you directly. Worth it only if you have a contract with the upstream.
  3. Models catalog (Models catalog). Pricing per million tokens, context window, features (tools, vision, JSON mode) and upstream health. Each model has a page such as https://anyrouter.dev/model/openai/gpt-5.4 and a markdown spec at /model/openai/gpt-5.4.md.
  4. Per-key limits (API keys, rate limits section). On the Keys page set requests-per-minute, tokens-per-minute or a daily spend cap, plus an expiry for short projects. A good production default is a daily cap at 2x expected usage and a generous RPM.

Next

API overview, Chat completions, Responses API, Errors. Migrating: from OpenAI, from Anthropic, to Responses.

AnyRouter routes requests to third-party models. Models and outputs are provided as is, with no warranty; you use them at your own risk. Disclaimer