LlamaParse

Send a document to the LlamaParse API for parsing. The node uploads the input file, polls the parse job until it completes (or the configured timeout elapses, failing the step), then fetches the result with the configured expanded fields and emits it as a JSON file on the response output. A job that ends as FAILED or CANCELLED fails the step with the provider's error message.

Pre-requisite: Install a LlamaIndex application in Profile > {Organization} > Applications (API key).

Parameters

Providerrequired

Configured LlamaIndex application.

Parse configuration

JSON object passed through untouched as the configuration field of the LlamaParse upload call, so any option the API accepts works here. New nodes start with {"tier": "agentic", "version": "latest"}. Clear it to fall back on the API defaults.

Expanded fields

Comma-separated fields fetched once the job completes (the expand query parameter). Defaults to markdown,items,metadata when left empty.

Timeout (seconds)

How long to wait for the parse job before failing the step. Defaults to 600. Large documents on slower tiers need a higher value.

Input

Filerequired

The document to parse.

Output

Response

A JSON File containing the completed parse job result with the configured expanded fields (by default the markdown rendition, parsed items and metadata).