# `AI_APICallError: 429 Too Many Requests — Rate limit exceeded (Retry-After: …)`

**Where it appears:** Terminal or Vercel runtime logs

**What it means:** Too many AI requests were sent in a short burst. Tokentide refused the extra request without charging and included a `Retry-After` header saying how many seconds to use the backup.

## Likely causes

- A user or automated test triggered many AI requests rapidly without a rate limit on your app's endpoint.
- The AI client retried Tokentide immediately instead of honoring `Retry-After` and routing to the Vercel AI Gateway fallback.

## Message for your agent

```text
We hit `429 Too Many Requests` on AI calls. Confirm our Tokentide wrapper sets `maxRetries: 0`, falls back to the Vercel AI Gateway while honoring `Retry-After`, and add a rate limit to our AI endpoint.
```

**Prevention:** Always add rate limits to model-calling endpoints and honor `Retry-After` in the Tokentide-to-Gateway fallback.

---
Source: [Agentic Makers](https://agenticmakers.app/en/help/errors/tokentide-429-rate-limit-exceeded) · License: CC BY 4.0