Can Mistral AI turn a simple, clearly formatted table into clean Markdown?
Mistral AI preserved a simple, clearly formatted table: the returned Markdown kept the four headers in order, retained eight data rows, and matched all 32 data cells against the source. The table stayed a single pipe table, and the comma inside "Metals suite, eleven elements" did not split the cell.
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.
Simple table — structure retainedPassEvidence
“schedule_of_analysis_charges_2026.pdf, 'the complete PDF, 3 page(s), in one request'”
# Kirkhollow Analytical Laboratories Ltd Braithe Road, Vardenne BR7 2HD 01632 960118 · reception@kirkhollow-labs.example ## Schedule of Analysis Charges Applicable from 1 April 2026 · Supersedes the schedule of 1 April 2025 The charges below apply to samples submitted under an open account. They are per sample and exclude value added tax. Turnaround is counted in working days from booking in, not from delivery. | Determination | Method | Turnaround | Charge per sample | | --- | --- | --- | --- | | Total suspended solids | KAL-M04 | 2 days | £18.40 | | Chloride | KAL-M11 | 3 days | £21.75 | | Ammoniacal nitrogen | KAL-M12 | 4 days | £26.50 | | Total oxidised nitrogen | KAL-M13 | 5 days | £29.20 | | Orthophosphate | KAL-M15 | 6 days | £31.60 | | Dissolved organic carbon | KAL-M22 | 7 days | £44.05 | | Metals suite, eleven elements | KAL-M31 | 8 days | £96.80 | | Particle size distribution | KAL-M47 | 10 days | £58.35 | Samples that require digestion before analysis are charged at the rate shown plus £9.15. A minimum charge of £75.00 applies to a single submission. Determinations not listed are quoted on request. ### How to submit samples Every submission needs a completed submission form, which is available from reception, from the sampling officer, or as a download. The form asks for the client reference, the sampling point, the date and time of sampling, the determinations required and a contact for the report. A submission that arrives without a form is booked in as received but is not analysed until the form follows, and the turnaround runs from the day the form arrives. Samples may be delivered to reception between 08:00 and 17:30 on weekdays and between 09:00 and 12:00 on Saturdays. Clients who hold a key may use the refrigerated drop box at the rear entrance outside those hours. Anything left elsewhere on the site is not accepted, and anything left in the drop box over a weekend is booked in on the Monday. ### Containers and preservation Containers are supplied without charge in the quantities a client's usual programme requires, and are collected on the same round wherever that is practical. Containers for metals are acid-washed and must not be rinsed at the sampling point. Containers for microbiology are sterile and single-use. Samples arriving in containers this laboratory did not supply are accepted at the client's risk, and the report carries a note to that effect. Preservation up to the point of delivery is the client's responsibility. Samples for suspended solids and for particle size need no preservative but should be kept cool and delivered within twenty-four hours. Samples for ammoniacal nitrogen should be chilled from the moment of sampling; a result on a sample that has not been is reported with a note saying so. Kirkhollow Analytical Laboratories Ltd · Registered in England 2214887 Page 1 of 3 Schedule of Analysis Charges Applicable from 1 April 2026 ## Turnaround and priority work The turnaround shown against each determination is the normal one and is counted in working days from booking in. It is the time to the issue of the report and not the time to the analysis, which is usually a day or two earlier. A submission arriving after 15:00 is booked in on the following working day. Priority work is accepted where the laboratory has capacity. A submission marked priority is booked in ahead of the queue and reported in half the normal turnaround, rounded up to the next working day, and is charged at the rate shown plus fifty per cent. Priority cannot be arranged retrospectively: a submission already in the queue keeps its place. ## Reports Reports are issued by e-mail to the contact named on the submission form, with a copy to the account address where the two differ. A report is issued once. A corrected report, where one is needed, is issued as a new document with its own number and says on its face what it replaces and why.
Returned Markdown contains one pipe table with matching headers and eight rows. From outside the graded run; shown for context, not graded.
Copied from Output file (Markdown), lines 1–52 · an excerpt
Open the whole Markdown file ↗These are scenario-level criteria. Each test case's Expected and Found are listed separately.
- Whether the rows and columns remain aligned after processing.
- Whether each value stays under the correct header.
- Whether the output still reads as a table rather than a flattened or split block of text.
| ✓ | Found: The table stayed intact with matching headers, eight rows, and all 32 values. |
The returned Markdown kept the source table intact: the same four headers, eight data rows, and all 32 cell values matched. The two tricky cells also stayed whole, including the comma in “Metals suite, eleven elements” and “10 days” in the turnaround column.
No ruling lines. The source table has no ruling lines, so column membership depends on reading order rather than visible grid lines.
Double-numbered lists. Ordered lists were double-numbered in the output; the repeated markers were output tokens, not source content.
Page furniture separate. Page furniture stayed outside the table body and appeared as separate paragraphs.
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 | Table Extraction → | |
| Scenario | A simple, clearly formatted table → | |
| 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.