AnyRouter Docs

Free-tier Aggregation

Combine several providers' free tiers behind one preset endpoint with key balancing, cooldown and an opt-in paid fallback.

Bring free keys from Groq, Cerebras, SambaNova, NVIDIA, Mistral, OpenRouter, GitHub Models, Cohere, Z.ai and others, group them in a preset, and call it like any model. Each request goes to a key that still has budget; a 429 puts that key on cooldown.

In Dashboard → BYOK, click Add key, pick a provider, paste the key, add a label and click Validate & Add. AnyRouter validates the key and shows detected limits:

  • Z.ai: 5-hour token window plus per-minute request rate
  • Anthropic (Claude OAuth): 5-hour, 7-day and per-model windows
  • DeepSeek: USD credit balance
  • OpenRouter: USD credit balance and key rate limit
  • OpenAI: daily request and token counts (admin keys only)
  • Groq, Cerebras, SambaNova, NVIDIA, Mistral, Cohere, GitHub Models: per-minute request and token budget from response headers

If a provider exposes no usage data, the key is saved as Unvalidated; set limits manually under Advanced if you want.

Open Dashboard → Presets → New. A preset is a list of branches, each a (model, upstream) pair such as Z.ai Standard API, Z.ai Coding Plan, OpenRouter or AnyRouter platform credits. The same model can appear in two branches via different upstreams. Set a weight or priority per branch, then pick the mode:

  • Fallback: Branch 1 first; Branch 2 only when every key in Branch 1 is cooled down or exhausted.
  • Weighted: spread requests by weight (for example 70/20/10).
curl https://anyrouter.dev/api/v1/chat/completions \
  -H "Authorization: Bearer sk-ar-your-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "@presets/duyet-coding",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Embeddings work the same on /api/v1/embeddings with "model": "@presets/my-embeddings". The preset's kind must match the endpoint (chat or embedding), otherwise the API returns 400 preset_kind_mismatch.

How it works

The resolver picks a branch, then a key in it using your per-provider key strategy (round-robin, weighted, fallback or random, set on the BYOK page). A key that returns 429 is on a 120-second cooldown (longer if the provider sends Retry-After), and the next request tries the next key. When every key in every branch is exhausted the request returns 429, unless Paid fallback is on.

Paid fallback (per preset, off by default) adds a lowest-priority "anyrouter platform" branch, so exhausted BYOK branches are served on platform credits. Cost appears in the same cost_usd column in the dashboard.

Capacity card

Each preset page shows three meters: Tokens and Requests (daily budget minus used today, summed over the preset's BYOK keys) and Platform $ (spend vs budget when paid fallback is on). Each key shows one status:

  • ● healthy
  • ◐ cooldown (seconds of Retry-After left)
  • ○ exhausted (95% or more of daily budget used)
  • ✕ invalid (rejected on the last probe)

Countdowns show the next reset per window. Click Probe all keys to re-run every probe (3-second budget each), or probe one key with POST /api/v1/byok/probe and { "key_id": "byok_..." }. Probes run only when you ask, not on a schedule.

Not included

  • OAuth-subscription tiers (Claude Code, GitHub Copilot, Cursor), whose terms do not allow third-party intermediation.
  • Sticky sessions: every turn re-resolves the preset. Pin a single model if you need affinity.

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