Rate Limit Query

[lpmitra]

Hello, I am writing a small vocabulary app over the summer. I am running into the rate limit, but I still don't know the exact limit. The way my api is designed: You input the words you want to look up, and it gets all of them at once. Could you fill me in on what the rate limit is?

[Junaid-PK]

The limit configured in the current server source is 450 requests per 5-minute window per IP (limiter definition). The limiter is mounted globally before the version/language/word route (middleware order).

There is one important wrinkle: a 429 does not always mean that your client sent 450 requests. This API fetches definitions from an upstream source, and an upstream 429 is also converted into the API own rate-limit error (dictionary.js lines 156–160).

The endpoint accepts one word per request; there is no batch endpoint in the current route. For a vocabulary app, I would therefore:

• de-duplicate the requested words and cache successful definitions locally;
• use a small concurrency limit instead of firing the whole list simultaneously;
• retry 429 responses with exponential backoff, honoring Retry-After when the response includes it.

Staying below 450 requests per five minutes avoids the application limiter, but the cache and backoff are still necessary because the shared upstream can throttle independently.