Free Tier

anyrouter/free routes your request to the best free model available right now. No provider keys, no credit deduction, and no per-account daily cap.

Make a request

curl https://anyrouter.dev/api/v1/chat/completions \
  -H "Authorization: Bearer $ANYROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anyrouter/free",
    "messages": [{ "role": "user", "content": "Hello, what model am I talking to?" }]
  }'

The model field in the response names the free model that answered (also in Dashboard → Logs). The current candidates are at /model/anyrouter/free. If the first model is unavailable or rate-limited, the next candidate is used.

Limits

anyrouter/free and all :free models share one per-minute limit per account, however many API keys you use. Decision models have their own limit. Over the limit returns 429 rate_limit_exceeded with Retry-After in seconds. See Rate Limits.

For comparison, paid models allow 120-1800 req/min by plan and BYOK 600 req/min.

On the Free plan

A Free account gets 5 successful requests, once, across free and decision models. API keys and the signed-in app share them, failed or cancelled requests do not count, and each trial response carries x-anyrouter-free-requests-remaining. The Credits page shows how many are left.

After the trial, requests return 403 free_tier_requires_subscription, and error.actions lists the ways into Go:

• subscribe: Go for $2/mo.
• donate_key: donate one working provider key no other account has donated to the shared Pool. An active donated key gives you Go for free.

Buying credit does not unlock free models; on Free it pays for paid models only. Your own BYOK keys work as normal.

Creating API keys on the Free plan

To keep free models available to everyone, new keys on the Free plan are limited:

• Disposable email addresses, or a second account on the same inbox: 403 free_key_restricted.
• Free-plan accounts on an email domain with an unusual burst of new accounts in 24 hours cannot create keys until it passes: 403 free_key_restricted. Large providers such as Gmail, Outlook and iCloud are never affected.
• One network can create up to 3 Free-plan keys in 24 hours: 429 free_key_ip_limit.

None of this applies once you have topped up at least $2 in total, or on any paid plan. Existing keys keep working.

Behavior details

• Faster first token. On streaming requests, if the first free model is silent for about 2.5 seconds, one more free model starts in parallel; the first token wins and the other is cancelled, and only one request counts. See Parallel upstreams to turn it off.
• :free fallback. If every route for a :free model fails, anyrouter/free answers instead, with the header x-anyrouter-fallback: anyrouter/free. Pin another model or check the header if you need that exact model.
• Plain ids with no balance. On Go and up with no credits, a free model called by its plain id (<id>, not <id>:free) on Chat Completions, Messages or Responses is served as <id>:free at $0 with the same limits, and the response sets x-anyrouter-served-model: <id>:free. With credits it routes and bills as usual. If the request carries fallback models, a preset or a family fallback, you get a 402 whose error.free_alternative names the :free id to send.

More capacity

1. Subscribe. Go ($2/mo) adds $4 monthly credit; Pro ($10/mo) and up add larger credit and higher limits. See Credits & Billing.
2. Add credits to call paid models (Go and up).
3. Add BYOK keys on the BYOK page and use anyrouter/byok; BYOK never deducts credits. See Free capacity with your own keys.
4. Pin a :free model (for example meta-llama/llama-3.3-70b:free) on Go and up; it shares the same limits.