Cant get Llamaparse to work

Questions
1

I’m trying to extract text from an image using Llamaparse.
This is my workflow:

All seem to work fine, the image gets uploaded, but the output is just rubbish.
When I use the Llamaparse Playground with the same image, it gives a perfect result.

ss
ss1481×193 39.4 KB

Image seems to get uploaded, but nothing is happening with it, it only outputs some garbage.

Anyone has any idea what I am doing wrong?

Thx,
Peter

2

It looks like your topic is missing some important information. Could you provide the following if applicable.

  • n8n version:
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app):
  • Operating system:
3
  • **n8n version:**Latest
  • Database (default: SQLite): N/A
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app): Cloud
  • Operating system: WIN11
4

hello @peter_wett

try with HTTP Request node’s option Response > Format (json/text)
image

5

Hi,
Thanks for the reply. But it didn’t work. It only put the output in another format.

thx,
Peter

6

Hey @peter_wett

The LlamaCloud UI uses premium_mode as the default selected for “parsing setting” but the API doesn’t - see documentation here.

If your HTTP request node, you may want to add the body paramater premium_mode=true and try again to see if that improves the output.

7

Hi Jim,

thanks for the answer.
I put ‘premium_mode’ as true in the body, and that improved the result a little bit. But still gibberish.

I see that that image is received at Llamaparse, but nothing is done with it as it still uses 0 credits.

I been looking through other parameters and I cant see where I can say ‘now get the text out of the image’ :grinning:

8

Just gave this a closer look and it seems you’re using a mix of the old API (/api) with the newer API (/api/v1). This may likely be cause of the issue.

Here’s an updated template which I was able to confirm it matches the UI.

  • Switched all endpoints to /api/v1
  • added premium_mode=true to api/v1/upload
  • Fixed typo in markdown endpoint from result/markdown to result/raw/markdown
5 Likes
9

You’re a hero!

Everything seems to be working now.

Thanks a lot!!!

10

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.