Rate limit / 429 errors
Symptom
During a Bulk or One-Click run, generation slows down or fails with an "API error (429)" — the AI provider is rate-limiting your requests because too many are being sent in parallel.
Cause
AISA Turbo sends multiple requests in parallel (×2/×4/×8...) to optimize thousands of pages quickly. If your provider's per-minute limit is lower than the parallelism you chose, the provider returns HTTP 429 (too many requests).
Fix
Lower the Turbo / parallel speed, or add more API keys so the load is spread across them. Turbo also auto-tunes: when it detects 429s it backs off automatically. For very large sites, use DeepSeek for the text — its rate limits are generous. See the "Provider limits & rate-limit" box on the API keys page for each provider's specifics, and add keys there to raise your effective throughput.
Questions people ask
Will a 429 lose my progress? No. Turbo backs off and retries; the run continues, just slower until the limit clears.
Does adding a second key really help? Yes. Requests are spread across keys, multiplying your available rate.
Which provider is best for huge sites? DeepSeek for the body text, thanks to its generous limits and low cost.
Related
- "Empty response from provider" / connection errors
- Gemini errors (deprecated model / empty output)