If you are building an application that sends multiple requests to the Perplexity API at the same time, you need to know the concurrency limit. Sending too many parallel requests can return HTTP 429 Too Many Requests errors and interrupt your service. The concurrency limit controls how many API calls your account can make simultaneously. This article explains the exact limits for Free, Pro, and Enterprise plans and how to stay within them.
Key Takeaways: Perplexity API Concurrency Limits
- Free Plan: 1 request at a time. Any additional request returns a 429 error.
- Pro Plan: 5 concurrent requests. Suitable for small to medium automation tasks.
- Enterprise Plan: Custom limits negotiated with Perplexity. No fixed cap.
Why Perplexity Enforces a Concurrency Limit
The concurrency limit protects the API infrastructure from overload. When one account sends too many parallel requests, it can degrade response times for all users. Perplexity sets per-plan caps to ensure fair resource distribution.
The limit applies to the number of active requests at any given moment. A request is active from the moment the client sends the HTTP POST until the API returns a complete response. If you send 10 requests at once on a Pro plan, only 5 will start processing. The other 5 receive an immediate 429 error and are not queued.
How the Limit Differs From Rate Limits
Concurrency is not the same as a rate limit. A rate limit restricts requests per minute or per hour. Concurrency restricts simultaneous in-flight requests. For example, the Pro plan may allow 100 requests per minute but only 5 at the same time. You can send 5 requests, wait for them to finish, then send 5 more.
How to Check Your Current Concurrency Limit
- Log in to the Perplexity API Dashboard
Go to perplexity.ai/settings/api. Sign in with your account credentials. - Locate the Plan Details Section
Scroll to the section labeled Plan Details or Usage. Your plan name and concurrency limit appear next to Concurrent Requests. - Check the HTTP Response Headers
After a successful API call, inspect the response headers. Look forX-RateLimit-ConcurrentandX-RateLimit-Concurrent-Remaining. These headers show your limit and how many concurrent slots are still free.
What Happens When You Exceed the Concurrency Limit
When your account sends more parallel requests than allowed, the API returns an HTTP 429 status code. The response body contains a JSON object with an error message and a retry_after field. The retry_after value tells you how many seconds to wait before sending a new request.
Example response:
{
"error": {
"message": "Too many concurrent requests. Limit: 5.",
"type": "rate_limit_error",
"retry_after": 10
}
}
The API does not queue excess requests. You must implement retry logic in your application to handle these errors gracefully.
If You Need More Concurrency
Free Plan Users
The Free plan allows only 1 concurrent request. To increase concurrency, upgrade to the Pro plan. Go to Settings > Subscription and select the Pro tier. After payment is processed, your concurrency limit increases to 5.
Pro Plan Users
If 5 concurrent requests are not enough, contact Perplexity support through the dashboard. Explain your use case and expected volume. Perplexity may offer a custom Enterprise plan with a higher concurrency cap.
Enterprise Plan Users
Enterprise agreements include custom concurrency limits. Your account manager provides the exact number in the contract. You can also request a temporary increase for testing or launch events.
Common Issues With Concurrency Limits
429 Errors Even When Staying Under the Limit
If you receive 429 errors but believe you are within the concurrency limit, check for other rate limits. The API also enforces a per-minute request cap. You might be exceeding that cap even if concurrency is fine. Review the X-RateLimit-Remaining header to see your per-minute allowance.
Multiple API Keys From the Same Account
Using multiple API keys from the same account does not increase concurrency. The limit applies to the account, not the key. All keys share the same concurrency pool. Generate separate accounts for separate applications if you need more concurrency.
Concurrency Limit Resets After a Long Request
A single long-running request (for example, a complex search that takes 30 seconds) occupies one concurrent slot for the entire duration. Other requests must wait. To avoid blocking, set a timeout on your HTTP client. If the request exceeds a reasonable time, cancel it and retry.
Perplexity API Free vs Pro: Concurrency and Other Limits
| Item | Free Plan | Pro Plan |
|---|---|---|
| Concurrent requests | 1 | 5 |
| Requests per minute | 20 | 100 |
| Models available | Perplexity Online only | Perplexity Online, GPT-4, Claude 3 |
| Monthly cost | Free | $20 per user per month |
Enterprise plan limits are not listed because they are negotiated per contract. Contact Perplexity sales for details.
Now you know the concurrency limits for each Perplexity API plan. Check your plan in the API dashboard and monitor the X-RateLimit-Concurrent-Remaining header in your app. If you need more concurrency, upgrade to Pro or request an Enterprise plan. Use a queue system in your code to send requests one batch at a time and avoid 429 errors.