Does Extend AI read multi-column pages in the right order?
Extend AI preserved the reading order of a two-column bulletin. It returned each page’s left column before its right column, without interleaving or inventing words.
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.
Two-column page — natural reading orderPassEvidence
The summer excursion will be to Marnhead on Saturday 20 June, to see the salt pans and what is left of the tramway on the north side of the harbour. Mr Tarbert has agreed to lead the walk, which is about three miles over rough ground and is not suitable for anyone who cannot manage a stile. The coach leaves the Assembly Rooms at half past nine and is back by six. The cost, including the coach and the entry to the harbour office, is fourteen pounds, and members should book with the excursion secretary before 1 June.
Returned Markdown: the converted bulletin
Copied from Output file (Markdown), lines 23–25
Open the Markdown file ↗[ { "object": "chunk", "id": "chunk_1_nld9PQ", "type": "page", "content": "The Vellamy Society\n\nBulletin\n\nFounded 1898\n\nNumber 212 . March 2026\n\n### Spring lecture programme\n\nThree lectures are arranged for the spring, all at the Assembly Rooms on Vardenne Quay at half past seven. On 11 March Dr Aileen Prosser wil … [4,198 characters in the file]", "metadata": { "pageRange": "{… 2 fields in the file}", "minOcrConfidence": 0.551, "avgOcrConfidence": 0.994 }, "blocks": [ "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}" ] }, { "object": "chunk", "id": "chunk_2_NPMniL", "type": "page", "content": "The Vellamy Society . Number 212 . March 2026\n\nThe committee's view remains that the interest of the place is the winch house and the slipway rather than the shed itself, which was re-roofed in the 1960s and has lost its doors. A measured s … [4,570 characters in the file]", "metadata": { "pageRange": "{… 2 fields in the file}", "minOcrConfidence": 0.319, "avgOcrConfidence": 0.994 }, "blocks": [ "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}" ] }, { "object": "chunk", "id": "chunk_3_W0fGwm", "type": "page", "content": "The Vellamy Society . Number 212 . March 2026\n\nroom hire from the reserve rather than from subscriptions, and to review that at the annual general meeting.\n\n### Reviews\n\nA History of the Marnhead Salt Trade, by J. Cavenshaw (Vardenne, 2025, … [3,621 characters in the file]", "metadata": { "pageRange": "{… 2 fields in the file}", "minOcrConfidence": 0.505, "avgOcrConfidence": 0.994 }, "blocks": [ "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}", "{… 8 fields in the file}" ] } ]
Excerpt ends here. Open the source for the complete output.
API response: conversion status and three page-by-page output chunks
Copied from Output file (JSON), parse_run › output › chunks · an excerpt
Open the whole JSON file ↗| ✓ | Expected: Text appears in natural reading order without interleaving the columns.Found: The Markdown put each page’s left column before its right column, with no interleaving or invented words. |
The conversion needed to follow the columns in natural order. It did so on each page without interleaving; the changed masthead punctuation did not affect the reading order.
Masthead punctuation. In the bulletin’s issue and date line, Extend AI changed the middle dot to a full stop.
Sentence across columns. On page 1, the sentence about the coach crossing the column break remained in order but was split into two paragraphs.
Page number placement. Extend AI tagged the page 1 number on its own line after the correspondence footer, rather than merging it into the body text.
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 | Reading Order & Layout → | |
| Scenario | A page laid out in multiple columns → | |
| Tool | Extend AI → |
Global scenario definition → · Global capability definition → · Extend AI 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.