Can Extend AI extract text accurately from a cleanly scanned document?
Extend AI passed this scenario. It returned the same three values from the digital invoice and the scanned twin, and the twin file was image-only. The digital run and the scanned run are shown separately, with different file IDs and matching invoice details: total due £1,754.40, invoice date 10 March 2026, and invoice number TPH-3301.
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.
Image-only twin, same schemaPassEvidence
{ "value": { "total_due": "£1,754.40", "invoice_date": "10 March 2026", "invoice_number": "TPH-3301" }, "metadata": { "total_due": { "citations": [ { "page": 2, "fileId": "file_Zj5mK2IV6kri5CM8OwJn5", "polygon": [ { "x": 482.6448000000001, "y": 143.4528 }, { "x": 533.3976, "y": 143.4528 }, { "x": 533.3976, "y": 154.40399999999997 }, { "x": 482.6448000000001, "y": 154.40399999999997 } ], "pageWidth": 594, "pageHeight": 841.0032, "referenceText": "£1,754.40" } ], "ocrConfidence": 0.977 }, "invoice_date": { "citations": [ { "page": 1, "fileId": "file_Zj5mK2IV6kri5CM8OwJn5", "polygon": [ { "x": 485.2008, "y": 165.35520000000002 }, { "x": 543.9816, "y": 164.988 }, { "x": 543.9816, "y": 174.4776 }, { "x": 485.2008, "y": 174.4776 } ], "pageWidth": 594, "pageHeight": 841.0032, "referenceText": "10 March 2026" } ], "ocrConfidence": 0.985 }, "invoice_number": { "citations": [ { "page": 1, "fileId": "file_Zj5mK2IV6kri5CM8OwJn5", "polygon": [ { "x": 484.1064, "y": 148.9248 }, { "x": 524.9952, "y": 148.9248 }, { "x": 524.9952, "y": 161.3376 }, { "x": 484.1064, "y": 161.3376 } ], "pageWidth": 594, "pageHeight": 841.0032, "referenceText": "TPH-3301" } ], "ocrConfidence": 0.994 } } }
{ "value": { "total_due": "£1,754.40", "invoice_date": "10 March 2026", "invoice_number": "TPH-3301" }, "metadata": { "total_due": { "citations": [ { "page": 2, "fileId": "file_1tciVugAPO7moc8RcxlrE", "polygon": [ { "x": 483.37919999999997, "y": 144.17999999999998 }, { "x": 535.5864, "y": 144.17999999999998 }, { "x": 535.5864, "y": 155.86559999999997 }, { "x": 483.37919999999997, "y": 155.86559999999997 } ], "pageWidth": 594, "pageHeight": 841.0032, "referenceText": "£1,754.40" } ], "ocrConfidence": 0.985 }, "invoice_date": { "citations": [ { "page": 1, "fileId": "file_1tciVugAPO7moc8RcxlrE", "polygon": [ { "x": 488.8512, "y": 167.9112 }, { "x": 546.1704, "y": 168.2712 }, { "x": 546.1704, "y": 175.9392 }, { "x": 488.8512, "y": 175.572 } ], "pageWidth": 594, "pageHeight": 841.0032, "referenceText": "10 March 2026" } ], "ocrConfidence": 0.989 }, "invoice_number": { "citations": [ { "page": 1, "fileId": "file_1tciVugAPO7moc8RcxlrE", "polygon": [ { "x": 488.12399999999997, "y": 153.6696 }, { "x": 526.4567999999999, "y": 153.30960000000002 }, { "x": 526.4567999999999, "y": 161.3376 }, { "x": 488.12399999999997, "y": 161.3376 } ], "pageWidth": 594, "pageHeight": 841.0032, "referenceText": "TPH-3301" } ], "ocrConfidence": 0.993 } } }
| ✓ | Expected: The same values from both runs, with both results shown.Found: Both runs returned total due £1,754.40, invoice date 10 March 2026, and invoice number TPH-3301; the twin was image-only. |
Both runs were captured and both returned the same values. The twin was verified as image-only from the file itself, and the two runs were distinct files with matching invoice details.
OCR confidence on digital. The digital run also returned OCR confidence scores, so it was processed through OCR too.
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.
Other tools on this scenario
How the 10 tools in this benchmark did on this scenario. A tool with a published result links to it.
Extend AI on the other scenarios
19 scenarios in this benchmark. A scenario with a published result links to it.
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 | Structured Document Extraction | 19 scenarios · 10 tools |
| Capability | OCR | |
| Scenario | A cleanly scanned document | |
| Tool | Extend AI |
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.