Can Mistral AI read a page laid out in multiple columns in the right order?
Mistral AI read the two-column page down each column in turn rather than across the page. On the three pages, the left and right columns stayed in order without interleaving.
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 counts this row.
Two-column page — natural reading orderPassEvidence
# The Vellamy Society *Bulletin* Founded 1898 Number 212 · March 2026 ## Spring lecture programme Three lectures are arranged for the spring, all at the Assembly Rooms on Vardenne Quay at half past seven. On 11 March Dr Aileen Prosser will speak on the salt industry of the upper estuary; on 8 April Mr Colm Tarbert will describe the 1953 flood as it was recorded at Kellow; and on 13 May the Society's archivist, Miss Ruth Sandiman, will introduce the Harrowby field books, which the Society acquired last autumn. Members may bring one guest without charge. Tea is served from seven o'clock, and the room is open from half past six for those who wish to look at the exhibits. Anyone needing a lift from the town should telephone the honorary secretary, who keeps a list of members willing to offer one. ## The Harrowby field books The eleven field books kept by Thomas Harrowby between 1871 and 1904 came to the Society in October, the gift of his great-granddaughter, and are now with the county record office for conservation. They are working notebooks rather than journals: tide times, wind, the state of the banks at Thrimby, the names of the men who worked each reach, and here and there a sketch of a channel that has since moved half a mile. The eighth book, which covers the winter of 1889, is the one most often asked for, because it records the breach at Kellow in some detail and gives the height of the water against the wall of the old lifeboat house. The Society hopes to publish a transcript of the eighth book, and would welcome help with it. What is wanted is not palaeography - the hand is a clear one - but local knowledge: field names, the older names for the reaches, and the identification of men named only by a surname and an initial. Members who could give an evening a month to this should write to the archivist. ## The summer excursion 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. ## New members and gifts The Society welcomes seven new members since the autumn bulletin: three from Vardenne, two from Braithe, one from Kellow and one, to the committee's particular pleasure, from Winnipeg. The library has received the bound volumes of the Braithe Advertiser for 1928 to 1934 from the estate of the late Mr E. Lissamer, and an album of photographs of the 1947 regatta from Mrs Prosser of Thrimby. Both have been accessioned and may be consulted on the usual notice. ## Notes and queries A member asks whether anyone can identify the building in the postcard reproduced in the autumn bulletin. The committee's view is that it is the old ferry office at Vardenne, pulled down about 1961, but the roof line does not match the photograph in the Society's own collection and the question is left open. Replies to the archivist, please, and not to the editor, who is away until April. ## Subscriptions Subscriptions for the year beginning 1 April are unchanged: eighteen pounds for a member, twenty-six pounds for two members at one address, and six pounds for members under twenty-five. Those who pay by standing order need do nothing. The Society remains grateful to the twenty-one members who add a donation to their subscription. ## The old lifeboat house
Markdown output with 133 lines and ordered gutter-spanning sentences. From outside the graded run; shown for context, not graded.
Copied from Output file (Markdown), lines 1–39 · an excerpt
Open the whole Markdown file ↗These are scenario-level criteria. Each test case's Expected and Found are listed separately.
- Whether text is emitted in the natural reading order down each column.
- Whether content from different columns is kept separate rather than interleaved.
- Whether the page's reading sequence is preserved even when all text is present.
| ✓ | Found: Natural reading order is preserved, with each left-column fragment followed by its matching right-column fragment. |
On all three pages, each gutter-spanning sentence stays in order, with the left-column text followed by the right. The cited output shows no column interleaving.
Page-break furniture split. At a page break, the footer, page number, and running header appear between two halves of one sentence. The sentence order still matches the requested reading order.
Gutter splits stay ordered. Sentences crossing a gutter are split into two paragraphs instead of stitched into one. Their order stays consistent at all three gutters.
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 row says which 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 row 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 | Mistral AI → |
Global scenario definition → · Global capability definition → · Mistral 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.