Still happening… and support emails are answered by an AI Agent! @Prashant_Pardesi
5 days with an outage is a serious matter; this is affecting the business
Hello, We just had our team look into your requests. Please respond to them via the support request and we should be able to identify the root cause. Thank you!
1 Like
@OpenAI_Support @Prashant_Pardesi
I just sent an email to support@openai.com regarding a vector store ID that is experiencing problems.
2 Likes
Any update from OpenAI on your side?
For us, we still experience a lot of “hanging” files when assigning them to a vector store. They get stuck “in progress” for more than 10 minutes where we let go.
Sometimes the vector store files moves on to completed after half an hour or more… or not at all.
We do have successfull assignments as well so the behaviour seems to be intermittent.
1 Like
Also experiencing this, stuck “in progress” status of vector stores
We have the same problem again
Openai needs a fail whale 
Will Write a massage to you in a bit ,also sharing the vectordtore - id.
Havent heard from anyone so far.
I’m talking to support, but honestly it’s not going anywhere for now.
Still happening btw. Seems to be intermittent, but it’s annoying—almost 1 week with that.
I’m looking actively for alternatives, this is happening every 2-3 months, 5 days of outage or more!
2 Likes
I feel you. We have already more than 1.000 documents that needs urgent processing…
2 Likes
This looks like a service-side issue with Vector Stores, not a client or SDK problem.
Files are accepted and processing starts, but the pipeline never completes and the vector_store_id is not returned, leaving the job stuck in in_progress until timeout. Multiple reports with the same behavior strongly suggest a backend regression.
From a practical standpoint, it’s risky to rely on Vector Stores as a single source of truth right now. A more resilient approach is to:
-
Store files and structured data in your own database/storage (DB, S3, Firestore, etc.).
-
Treat Vector Stores as an optional or best-effort semantic search layer.
-
Add a hard timeout and fail fast if
vector_store_idis not returned. -
Allow the system to continue operating without vectors while the service is unstable.
This way, when the Vector Store bug is fixed, you can re-enable it without blocking or breaking the rest of your application.
Well we need the vector stores to give chatgpt access to the file and analyze it according to our needs. We cant upload files directly to a gpt and start the analysis, can we?
Tried sending support details and keep getting sent in circles that we should just expect long delays in processing. Why in the world should vectorizing a few thousand String tokens from a pdf take more than 20 minutes? There is no AI going here, this is just fast math that should be almost instantaneous.
It’s still happening after 1 week! @Prashant_Pardesi @vb @OpenAI_Support
I’m not the only one here, I think you have to start working on this and fix it or give some information
I just started getting new behavior in the last few hours, after having super delayed processing all week…now the same file fails immediately to be put into the vector store, but the file was successfully uploaded. Weird how this API can be so non-deterministic.
Seems to be very intermittent, but it’s annoying, and the clients are complaining. The last one happened 1 hour ago in my case. So I guess it’s still active
1 Like
I pinged the team again in the hopes of getting this resolved. But unfortunately I have no timeline to share at this point in time.
Hoping this gets fixed for you quickly!
Seems that it’s fixed now.
But frankly, the OpenAI support in this case is terrible. NO email answers or not relevant, 0 information, and the status page doesn’t reflect anything about the 1-week outage on Vector Store.
The OpenAI API’s confidence went down radically
1 Like