Can LlamaParse use a supplied definition to choose the right value?
LlamaParse returned the consignment reference CFS-CN-40821, as the field definition asked, rather than the invoice number CFS-7710. The invoice number is printed more prominently in the header and on every page.
1 of 1 test case passed
Every test case in this scenario has a result.
The test case
Each test case is judged on its own: Pass, Fail, Not gradable, or Untested. The scenario result above is based on this test case.
Definition overrides the prominent numberPassEvidence
We uploaded the freight invoice to the Extract page with a schema defining one required text field whose description selects the consignment reference and excludes the invoice number.
{ "type": "object", "properties": { "reference_number": { "type": "string", "description": "The consignment reference number, the identifier tied to the physical shipment. This is NOT the invoice number, even though the invoice number appears more prominently (labelled, and repeated in the footer of every page)." } }, "required": ["reference_number"] }
"reference_number": "CFS-CN-40821"
Extraction output: the selected reference number.
Copied from Output file (JSON), lines 9–11 · an excerpt
Open the whole JSON file ↗Source excerpt · partial file · response 1
"extract_result": { "reference_number": "CFS-CN-40821" },
Copied from Output file (JSON), lines 9–11 · an excerpt
| ✓ | Expected: Return the value named by the supplied definition, not the more prominent one.Found: It returned the consignment reference on page 1 and did not return the invoice number. |
The field definition excluded the invoice number, and the returned value was the consignment reference on page 1. The repeated invoice number remained the more prominent alternative in the pages' headers and footers, so the definition controlled the answer.
Location and confidence. The consignment reference has a page-1 citation with a bounding box and confidence of 0.9501241833995158.
Configuration and setup
How this tool was set up for the run and what the test needed in place. Each row is a fact from the run's records; a fact the records do not hold is left out, not guessed.
How this scenario is graded
How we decide Pass, Fail and Not gradable. The same rules apply to every tool tested on this scenario.
Each test case gets one result per tool: Pass, Fail or Not gradable. A test case we haven't run yet shows Untested. There are no partial results.
- Pass: the tool did everything the test expected, and nothing it said contradicts the correct answer.
- Fail: at least one expected behaviour clearly didn't happen; the test case's result says which one and quotes the tool.
- Not gradable: our evidence couldn't settle the outcome (for example a record we needed is missing). It is never counted as a fail, and the test case's result says what's missing.
Where this sits in the benchmark
This page is one cell of a larger study: one tool, one scenario. Only this benchmark's frame appears here.
| Level | Name | Scope |
|---|---|---|
| Benchmark | Structured Document Extraction → | 19 scenarios · 10 tools |
| Capability | Field Extraction → | |
| Scenario | The value needs a supplied definition → | |
| Tool | LlamaParse → |
Global scenario definition → · Global capability definition → · LlamaParse product page →
History of this result
What has happened to this result since it was first published. Runs and grades are never overwritten: a retest or a re-grade publishes a new result and keeps the earlier one readable.
Act on this result
Nothing filed here edits the run or the grade. A challenge opens a review, and a review can produce a new run or a re-grade — which becomes the current result and leaves this one in the history.
You run the same kind of test against your own setup and get the same behaviour.
Agree →Yours behaves differently. Tell us what you got, with a screenshot if you have one.
Disagree →Something here is wrong — a reference value, a transcription, a grade.
Report an issue →Tell us what changed and we schedule a rerun of the failing test case. The old result stays as history.
Vendor notice →The same record is available as structured data through the AI Demos MCP server, with the counts, the coverage and every per-test-case reason carried as fields.
Verify the proof files
These files support this result. Open a file to inspect the original evidence.
File fingerprints (SHA-256)
A fingerprint identifies the exact file used for this result.