Perplexity (Sonar) models hitting upstream rate limits

[pollinations-ai]

Issue Summary

Perplexity models (perplexity-fast, perplexity-reasoning) are experiencing upstream rate limiting from Perplexity's API, causing requests to fail even for users with sufficient credits and tier limits.

Context

Reported by: cypherpotato (Spore tier, $50+ in credits)

User is experiencing rate limit errors when using Perplexity models. This is not a Pollinations tier/credit issue - the user has ample credits and is on Spore tier. The rate limiting is happening upstream at Perplexity's API level, likely due to overall volume across all Pollinations users hitting Perplexity's API key limits.

Impact

• Users cannot reliably use perplexity-fast or perplexity-reasoning models for batch processing
• Workaround (gemini-search) provides different result quality that doesn't meet user's needs
• User is currently having to manually queue/batch requests to work around the issue

Possible Solutions

1. Request higher rate limits from Perplexity for our API key
2. Implement retry logic with exponential backoff when hitting 429 errors
3. Add request queuing/batching on Pollinations side to smooth out traffic
4. Multiple API keys with load balancing across them

Related

• User is using these models for batch processing workflows where Gemini results don't match their requirements
• Issue affects all Perplexity models regardless of user tier or credit balance

---

Author: cypherpotato (UID: 226860943651504129)
Source: View on Discord

[ElliotEtag]

Closing as stale — Polly-generated 3+ months ago with no human follow-up. Please open a new issue with current details (endpoint, model, request, response body, timestamp) if this is still happening.