API very slow / returning 5xx errors

[wisnewskirobert]

hi there,
I am testing the perplexity API to see if I can use it in a new app.

It is taking around 47 seconds to return a response. Out of my last 6 tests I got 3 5xx errors and 3 200s.

1. is it normal to get that many 500 errors?
2. is this latency normal?

Not sure I will be able to use it with this kind of performance. Maybe its just a 1 off today?

Here are the details of my call:
Company Information: { "id": "e0b0e4d3-1ffb-402e-906a-3d6540216208", "model": "llama-3.1-sonar-huge-128k-online", "created": 1728991890, "usage": { "prompt_tokens": 128, "completion_tokens": 741, "total_tokens": 869 }, "object": "chat.completion",

thanks,

[wisnewskirobert]

Also here are my parameters:

max_tokens: 1000, // Adjust as needed
temperature: 0.2,
top_p: 0.9,
return_citations: true,
search_domain_filter: ["https://fr.indeed.com/", "https://www.glassdoor.fr/com/", "https://www.linkedin.com/"],
return_images: false,
return_related_questions: false,
search_recency_filter: "month",
top_k: 0,
stream: false,
presence_penalty: 0,
frequency_penalty: 1

[aarashy]

Hey, I recommend using sonar-large instead; it is much more stable.