Form Parser imageless mode still enforcing 15-page limit despite imagelessMode: true

Google Cloud AI Infrastructure
1

I’m hitting the following error on a Form Parser processor when processing documents with 16–24 pages via the synchronous process endpoint:

“This processor version only supports up to 15 pages per document shard. If using the UI or sync prediction API, this problem can be resolved by using the batch prediction API instead.”

Per the documented limits, Form Parser should support up to 30 pages synchronously when imagelessMode: true is set.

I’ve confirmed:

  • imagelessMode: true is set as a top-level field in the request body (sibling of rawDocument, not nested inside it — verified in n8n’s raw request payload)
  • Tested against both pretrained-form-parser-v2.0-2022-11-10 and pretrained-form-parser-v2.1-2023-06-26 (Release Candidate) — same error on both
  • Tested in both europe-west2 and eu (multiregion) processor instances — same error in both regions
  • Processor type confirmed as Form Parser via the processor details page

I’m on Basic support so can’t file a technical case directly. Is there a project-level allowlist required for imageless mode to actually take effect on Form Parser, similar to what’s mentioned for Custom Extractor? Any pointers appreciated.

Environment:

  • Processor type: Form Parser
  • Region: eu (multiregion)
  • API: v1/.../processors/{id}:process (synchronous)
1 Like