Cannot Parase Any Files

[Kodora-lab]

I've uploaded a series of PDF documents, many of which contain black and white text and are under 1MB each.

Not matter what settings I use, llama keeps timing out, or can't parse any file - yet still burns through process credits.

I've tried several file types. I would have thought uploading a single document and parsing would be easy but maybe there is a custom configuration or setting that's blocking the whole process.

[logan-markewich]

If you can share a job id or sample pdf, or any other details, that would help quite a bit with investigating/replicating.

What specifically is timing out? Are you calling through code or the UI? Any specific config?

[Kodora-lab]

I've attached a sample file for reference. I'm also using the UI directly.

[

1-2019_sporting_regulations_2018-12-05.pdf

](url)

[logan-markewich]

@Kodora-lab Any specific settings you were trying? I used that PDF with fast (text only, no markdown), cost effective, agentic, and agentic plus tiers and all 4 worked at first glance.

I guess something else to clarify, was this in cloud.eu.llamaindex.ai or cloud.llamaindex.ai ? There are some performance differences between the two regions due to model availability. I was testing on cloud.llamaindex.ai. (If you have the Job IDs for any failed jobs, that helps as well)

Some output screenshot for reference, and a sample of the output text, markdown and output JSON attached

output.txt
output.json
output (2).md

[Kodora-lab]

I'm using https://cloud.llamaindex.ai

I've tried again with another document but get this error.

I've also tried the same files using another tool (Unstructured.io) and it works immediately but I can't get any file to parse in Llama Cloud. I have also used agent and agent plus but no luck. Can you drop or post any other settings you used?

[logan-markewich]

So this is the indexing feature, not parsing, got it.

My above tests were purely in the parse playground, not the index tool.

I can see if I can use the details here to track down the issue

[logan-markewich]

@Kodora-lab looks like openai embeddings timed out. What settings did you select when creating the index? Are you using managed embeddings or you provided your own key?