Can Adobe PDF Extract API handle an ordinary digital text document accurately?
Adobe PDF Extract API did not keep the document text unchanged. Its Extract output joined the running header to the document title and to a sentence continuing onto page 2; no words were missing or invented.
1 of 1 test case failed
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.
Ordinary digital text — complete and unchangedFailEvidence
# Braithe Estuary Partnership Sediment recharge at Kellow Sands: the 2025/26 placement
Markdown output: full document assembled from Adobe’s response
Copied from Output file (Markdown), line 1
Open the Markdown file ↗{ "Bounds": [70.86610412597656, 760.0906066894531, 496.98809814453125, 792.6679992675781], "Font": { "alt_family_name": "Calibri", "embedded": true, "encoding": "Custom", "family_name": "Calibri", "font_type": "TrueType", "italic": false, "monospaced": false, "name": "AAAAAA+Calibri-Bold", "subset": true, "weight": 700 }, "HasClip": false, "Lang": "en", "ObjectID": 204, "Page": 0, "Path": "//Document/Title", "Text": "Braithe Estuary Partnership Sediment recharge at Kellow Sands: the 2025/26 placement ", "TextSize": 17.0, "attributes": { "LineHeight": 16.875, "SpaceAfter": 10.75 } }
Extract response: Adobe’s structured document output
Copied from Output file (JSON), structuredData › elements 1
Open the whole JSON file ↗{ "Bounds": [70.86610412597656, 767.4553985595703, 178.05430603027344, 792.6679992675781], "Font": { "alt_family_name": "Calibri", "embedded": true, "encoding": "Custom", "family_name": "Calibri", "font_type": "TrueType", "italic": false, "monospaced": false, "name": "AAAAAA+Calibri", "subset": true, "weight": 400 }, "HasClip": false, "Lang": "en", "ObjectID": 177, "Page": 1, "Path": "//Document/P[9]", "Text": "Braithe Estuary Partnership have shown so far. ", "TextSize": 11.399993896484375, "attributes": { "LineHeight": 12 } }
Copied from Output file (JSON), structuredData › elements 14
| ✗ | Expected: All text present and unchanged, with nothing missing, garbled or invented.Found: The running header was joined to the title and to a sentence continuation on page 2. No words were missing or invented. |
All text had to be present and unchanged. Both joins are already in Adobe's own structured output (the title element and the paragraph that continues onto page 2), so they were not introduced when we assembled the Markdown. Joined header text is garbled text, so the result fails.
Heading levels. The assembled Markdown places the document title and all 24 section headings at the same heading level.
Page footers. No page footer lines appear in the assembled Markdown.
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 | Converting a complex PDF into clean Markdown with a hosted API → | 12 scenarios · 13 tools |
| Capability | Text Fidelity → | |
| Scenario | An ordinary digital text document → | |
| Tool | Adobe PDF Extract API → |
Global scenario definition → · Global capability definition →
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.