AI Demos Research — the structured-intelligence platform. Every verdict on these pages opens to the execution behind it.
Graded 16 November 2026

Does LlamaParse keep footnotes out of the body flow and attach them to their references?

LlamaParse kept all 13 footnotes out of the body text and tied each one to its reference. Each page's notes come after that page's body, every note on its own line, and every reference mark in the body is a superscript that matches a note, with every note matching a mark.

1 of 1 test case passed

Every test case in this scenario has a result.

Pass rate100%1 of 1 gradable
Coverage1 of 1test case with a result
1PassDid everything it was expected to do.
0FailNo test case failed.
0Not gradableEvery result here could be graded.
0UntestedEvery test case in this scenario has a result.
The pass rate is a summary. The evidence is the test case below: what we sent, what we checked, what the tool returned, and the proof.

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.

Footnote — out of body flow, associatedPassEvidence
What we sent

We sent the complete four-page PDF to LlamaParse's parse API in one request, with every setting left at its default.

InputDocument (PDF)
Input PDF: the four-page journal offprint with 13 numbered notes at the page feet
Open the PDF ↗
What the tool returned

Returned Markdown: superscript marks in the body, each page's notes below it

Copied from Output file (Markdown), lines 23–29

Markdown
<sup>1</sup> Braithe Estuary Order 1974, art. 6(2). The Order is reproduced in full in the appendix to R. Sandiman, *The Braithe Estuary: A Documentary History* (Vardenne, 1988), 211-24.

<sup>2</sup> Minutes of the Harbour Board, 14 May 1973 (Braithe Record Office, BHB/3/17), 4.

<sup>3</sup> The point was not argued in *Tarbert v Braithe Harbour Board* [1981] 2 EstLR 88, where both parties proceeded on the assumption that art. 6(2) applied.

<sup>4</sup> The Board's file on the 2011 consents (Braithe Record Office, BHB/11/4) contains three letters returned undelivered and a note of a telephone call to a solicitor in Winnipeg.
Open the Markdown file ↗

API response: the completed job and all four pages processed

Copied from Captured file (JSON), job › status

Text
COMPLETED
Open the whole JSON file ↗
Expected vs. Found
✓Expected: The footnote text is present, kept out of the body text, and tied to the reference mark that points to it.Found: All 13 notes are present, grouped on their own lines below their own page's body, and each matches a superscript mark earlier in the body, in both directions.
Supporting proof
Proof 1Screenshot
Page 1 of 4, source beside output: marks 1 to 4 and their notes below the body
Open original ↗
Proof 2Screenshot
Page 2 of 4, source beside output: notes 5 to 7 below the body
Open original ↗
Proof 3Screenshot
Page 3 of 4, source beside output: marks 8 to 10 and their notes after the body
Open original ↗
Proof 4Screenshot
Page 4 of 4, source beside output: marks 11 to 13 and their notes below the body
Open original ↗
Proof 5Screenshot
Run details card: the default setup, the complete PDF sent and the completed run
Open original ↗
Cited values2 values from this file
job › tier
agentic
job › name
croyde_1974_braithe_order_offprint.pdf
Why this result

The notes had to be present, kept out of the body text, and tied to their references. All three held: all 13 notes appear, each below its own page's body and never inside a paragraph, and all 13 marks and notes match each other.

Also observed in this test case

Running heads kept as lines. The running heads and page numbers of the source pages are kept, each on its own line at a page boundary, and none is spliced into a sentence or a note.

Run-together marks separated. Where the source's text runs a note mark into the word before it, after 'willing to consent' and 'the position of tenants', the output keeps each mark as its own superscript; no body word differs.

Tested by Chandresh Bisht · evidence dated 10 September 2026

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.

Software that produced the output
LlamaParse
Build or version
Parse API v2, tier agentic, version latest (build not exposed)
Plan or tier
free tier
Surface
API
Set up before the run
A four-page journal offprint: body text with superscript reference marks, and 13 numbered notes below a rule at each page's foot. The complete PDF was to be converted to Markdown in one default pass, with no setting tuned.
Tested
10 September 2026 · Chandresh Bisht

How this scenario is graded

How we decide Pass, Fail and Not gradable. The same rules apply to every tool tested on this scenario.

How results are decided

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.

The rules
  • 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.

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.

from the publication record
17 November 2026First publishedConverting a complex PDF into clean Markdown with a hosted API v1

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.

This matches what I see

You run the same kind of test against your own setup and get the same behaviour.

Agree →
This does not match

Yours behaves differently. Tell us what you got, with a screenshot if you have one.

Disagree →
Point out an issue

Something here is wrong — a reference value, a transcription, a grade.

Report an issue →
Request a retest

On a newer build, a larger dataset, or your own setup.

Request a retest →
We have fixed this

Tell us what changed and we schedule a rerun of the failing test case. The old result stays as history.

Vendor notice →
Filed against this evidenceNothing yet. Challenges, counter-evidence and fix notices appear here with their outcome, and stay on the page after they are resolved.
Cite this result
aidemos.com/benchmarks/pdf-to-markdown-apis/results/llamaparse/footnotes-at-the-bottom-of-the-page · 1 pass · 0 fail · coverage 1/1 · graded 2026-11-16

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.

Input · Document (PDF) ↗application/pdf · 152 KB
Output file (Markdown) ↗text/markdown · 11 KB
Proof 1 · Screenshot ↗image/png · 893 KB
Proof 2 · Screenshot ↗image/png · 606 KB
Proof 3 · Screenshot ↗image/png · 408 KB
Proof 4 · Screenshot ↗image/png · 331 KB
Captured file (JSON) ↗application/json · 25 KB
Proof 5 · Screenshot ↗image/png · 66 KB
File fingerprints (SHA-256)

A fingerprint identifies the exact file used for this result.

Input · Document (PDF)fce3c87776b6f1b8f34dcfb0f6a274b5954edcb3d33ff7ee84840dfe2afbc1ad
Output file (Markdown)198e1f04db2e6194b281a41bc4ed793946513841fd48f7d674fec96586745655
Proof 1 · Screenshot4a59bf46fa2f51c5dd222d6ef24624ad7fdbaac039bab3e7d7b8e3aa6af3b7de
Proof 2 · Screenshot9b9d8b8b4bd721335e752049be144e0093fb8c925234e347fb4c771193105989
Proof 3 · Screenshotcf23e564f3ffcdb9144df4ba9101336a5a7c4eeb77d32e4ad68574b4ef109edf
Proof 4 · Screenshota0ab7a4009a18de9a3091b2adee7d64a56f9d092e8e61e8703e5d0897d3cabf6
Captured file (JSON)ad26a6a5959fe11742a9c4b2d28aad7a78965bd0d93203143dcd5f70cd16604d
Proof 5 · Screenshotbb01452cd8b1bd962777ab4fdd3545df18eeb3e2d65dda74c3153caebaff5299