I'm facing a lot of failed API calls (Assistant and vector store)

API Bugs
1

Hi there,

I have a SaaS using the Assistant API and we are facing a lot of failure since yesterday.

One problem with the Assistant API with those messages :

  1. You’ve exceeded the rate limit, please slow down and try again later.
  2. The server had an error processing your request. Sorry about that! You can retry your request, or contact us through our help center at help.openai.com if you keep seeing this error. (Please include the request ID req_xxx in your email.

Also a problem with the vector storage :

Failed to create vector store.
#0 /home/customer/www/app.xxx.ai/public_html/vendor/openai-php/client/src/Transporters/HttpTransporter.php(57): OpenAI\Transporters\HttpTransporter->throwIfJsonError(Array, ‘{\n “error”: {\n…’)
#1 /home/customer/www/app.xxx.ai/public_html/vendor/openai-php/client/src/Resources/VectorStores.php(32): OpenAI\Transporters\HttpTransporter->requestObject(Object(OpenAI\ValueObjects\Transporter\Payload))
#2 /home/customer/www/app.xxx.ai/public_html/app/Services/AI/AssistantService.php(87): OpenAI\Resources\VectorStores->create(Array)
#3 /home/customer/www/app.xxx.ai/public_html/app/Http/Controllers/User/RevisionsController.php(139): App\Services\AI\AssistantService->createVectorStores(Object(Illuminate\Http\UploadedFile))

My organization is tier 5, so we should not have any problem related to rate limit at this point.

I feel like this problem is coming from OpenAI directly.

Is anyone else facing this ?

Thanks in advance,
Zak.

1 Like
2

A 429 status code (You’ve exceeded […]) is sometimes used to indicate that the server itself is globally limited, and may not necessarily reflect your personal limits.

I can’t say it’s for certain, but there have been a lot of recent issues with Assistants based on the forum posts.

Best bet is just to be patient. It’d be interesting to know where your calling host geographically resides.

I run Assistants in a Canadian location (fly io) and haven’t experienced any slowdowns. So it may be worthwhile to have another instance setup in other locations?

1 Like
3

Same here - timeouts for simple calls like listing vector store files. Definitely not a user limit thing… (calling from Europe)

1 Like
4

Thanks for the answer !

I’m from France, so we are calling from Europe.

The OpenAI status page shows no problems…

This is what DownDetector is showing regarding OpenAI in the last 24 hours.

Capture d’écran 2025-01-14 à 18.39.02
Capture d’écran 2025-01-14 à 18.39.021564×636 37.8 KB

5

Yesterday we had lot of problems with rate limit and assistant and today we are facing most of the issues with vector storage…

Calling from Europe also.

6

Also via the UI (Dashboard / Storage) it just hangs (“Loading…”)

1 Like
7

We had a lot of issues and down times with the Assistant/VectorStore APIs :frowning:
I guess we will have to do all the RAG stuff (chunking, embedding, vector DB) manually/locally after all and only use OpenAI for the final inference until they get the Assistant API out of Beta and actually working reliably…

But for now it seems to be working again…

2 Likes