{"observation":{"id":"d572c22d-d6ae-4a81-b848-53c180d00575","tool":"mistral-ai","tool_name":"Mistral AI","criterion":"complex-document-handling","criterion_name":"Complex Document Handling","criterion_definition":"Maintains quality across long, multi-section, and mixed-content documents without degradation.","criterion_evidence_type":"transformation","criterion_rank_role":"decisive","criterion_rank_role_reason":"The subject explicitly says complex PDF, so sustained quality across long, mixed-content documents is central to the tool’s ability to do the job. (3 of 3 judges)","scenario":"scanned-research-paper","scenario_name":"Scanned Research Paper","group_tag":"scanned-research-paper","scenario_description":"An image-only scanned research paper used to stress OCR and layout recovery in a multi-column academic document with figures, charts, tables, captions, and references.","modality":"pdf","input_text":null,"input_artifact_refs":[{"alt":null,"url":"https://d3epheqghktydj.cloudfront.net/convert-a-complex-pdf-into-clean-markdow-scanned-research-pdf-7b86de49784d.pdf","role":"input","filename":"Scanned Research PDF.pdf"}],"stresses":["OCR on scanned pages","Multi-column reading order","Figure and chart handling","Table reconstruction from scans","Caption association","Reference extraction","Overall document structure retention"],"verdict":"worked","score":null,"score_total":null,"note":"The tool processes a scanned multi-column research paper end-to-end and returns OCR text, tables, and embedded chart assets in page-wise markdown output.","evidence_state":"verified","source":null,"artifacts":[{"url":"https://cdn.futuresmart.ai/public/aidemos/f84ba080d3254487844fc0a3f51f1873.mp4?v=1","role":"context","alt":null},{"url":"https://d3epheqghktydj.cloudfront.net/research-media-mistral-ai-scanned-pdf-output-zip-c2a22ad51169.zip","role":"output","alt":null},{"url":"https://cdn.futuresmart.ai/public/aidemos/45e3533a31c246b29e0ec6aaa98438e4.pdf?v=1","role":"input","alt":null},{"url":"https://cdn.futuresmart.ai/public/aidemos/c627598b071444a3931982b63cc44737.mp4?v=1","role":"context","alt":null}],"run_id":"6e3160de-fe46-4b45-b071-72560b5c5d0e","study_title":"Convert a Complex PDF into Clean Markdown with an API","study_kind":"generation","research_task":"86b9h7t37","tested_at":null,"completeness":"input-and-output","input":{"state":"files","text":null,"files":[{"url":"https://d3epheqghktydj.cloudfront.net/convert-a-complex-pdf-into-clean-markdow-scanned-research-pdf-7b86de49784d.pdf","filename":"Scanned Research PDF.pdf","alt":"Scanned Research Paper","role":"input"}],"modality":"pdf","stresses":["OCR on scanned pages","Multi-column reading order","Figure and chart handling","Table reconstruction from scans","Caption association","Reference extraction","Overall document structure retention"]},"tool_page_slug":"mistral-ai","tool_url":"https://aidemos.com/tools/mistral-ai","permalink":"https://aidemos.com/evidence/d572c22d-d6ae-4a81-b848-53c180d00575","api_url":"https://ai.aidemos.com/v1/observations/d572c22d-d6ae-4a81-b848-53c180d00575"},"peers":[{"id":"32fab15d-0885-4287-94a3-67e0d1d00611","tool":"adobe-api","tool_name":"Adobe API","verdict":"struggled","score":null,"score_total":null,"note":"Requires splitting an oversized scanned paper into two PDFs before processing, because the web interface rejects uploads over 1 MB.","artifact_count":6,"thumbnail":"https://cdn.futuresmart.ai/public/aidemos/790f9221306f4bd98ec30e9c44908862.mp4?v=1","evidence_url":"https://aidemos.com/evidence/32fab15d-0885-4287-94a3-67e0d1d00611"},{"id":"c8ea9121-f878-4e78-bf13-3ba3b4b6f0a5","tool":"extend-ai","tool_name":"Extend AI","verdict":"worked","score":null,"score_total":null,"note":"Handles the scanned research paper end-to-end, including multi-column prose, tables, charts, and handwritten marginalia, while still producing parsed markdown.","artifact_count":3,"thumbnail":"https://cdn.futuresmart.ai/public/aidemos/45e3533a31c246b29e0ec6aaa98438e4.pdf?v=1","evidence_url":"https://aidemos.com/evidence/c8ea9121-f878-4e78-bf13-3ba3b4b6f0a5"},{"id":"9f0d06be-7ecb-4732-933c-ea6d52a1cc16","tool":"llamaparse","tool_name":"LlamaParse","verdict":"worked","score":null,"score_total":null,"note":"Processes a 12-page scanned paper end-to-end and reaches SUCCESS after extracting the page content and figure/table outputs.","artifact_count":3,"thumbnail":"https://cdn.futuresmart.ai/public/aidemos/9848fddc3df54deda1f31e5ae8d86f56.mp4?v=1","evidence_url":"https://aidemos.com/evidence/9f0d06be-7ecb-4732-933c-ea6d52a1cc16"},{"id":"b50f09d7-7cee-4268-8441-50b46ddeb000","tool":"nutrient-io","tool_name":"Nutrient.io","verdict":"worked","score":null,"score_total":null,"note":"Processes an image-only scanned research paper end to end and returns a parsed markdown output plus preview, showing it can handle a multi-page scanned document.","artifact_count":4,"thumbnail":"https://cdn.futuresmart.ai/public/aidemos/45e3533a31c246b29e0ec6aaa98438e4.pdf?v=1","evidence_url":"https://aidemos.com/evidence/b50f09d7-7cee-4268-8441-50b46ddeb000"},{"id":"3bec7ed4-9b8e-44f3-b548-b76f0215b34c","tool":"reducto","tool_name":"Reducto","verdict":"struggled","score":null,"score_total":null,"note":"Handles the 12-page scan in 10.6 seconds with no truncation, but the densest 17-column table has unreadable stretches with injected glyphs and run-on numbers.","artifact_count":2,"thumbnail":"https://d3epheqghktydj.cloudfront.net/research-media-reducto-input3-scannedpaper-output-6dfdac76e722.md","evidence_url":"https://aidemos.com/evidence/3bec7ed4-9b8e-44f3-b548-b76f0215b34c"},{"id":"ea443095-bb7a-45ec-92ca-6ff0afefe8e4","tool":"tensorlake","tool_name":"Tensorlake","verdict":"mixed","score":null,"score_total":null,"note":"On the scanned research paper, section flow and chart extraction work, but hierarchical tables degrade, so mixed-content handling is uneven rather than consistently robust.","artifact_count":8,"thumbnail":"https://cdn.futuresmart.ai/public/aidemos/18b2277b09fa42e3ae7a325fba02b272.png?v=1","evidence_url":"https://aidemos.com/evidence/ea443095-bb7a-45ec-92ca-6ff0afefe8e4"}],"other_criteria":[{"id":"f0009fdb-31cf-4281-aceb-887be1538083","criterion":"markdown-quality","criterion_name":"Markdown Quality","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"The export is packaged as usable markdown files in a ZIP, with both overall and page-wise outputs available for inspection.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/f0009fdb-31cf-4281-aceb-887be1538083"},{"id":"cc8096bb-19af-4b8f-901f-657634e78a25","criterion":"reading-order-structure","criterion_name":"Reading Order & Structure","rank_role":"decisive","verdict":"failed","score":null,"score_total":null,"note":"The opening page loses the distinction between the document title and the abstract, flattening the semantic organization of the first page.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/cc8096bb-19af-4b8f-901f-657634e78a25"},{"id":"a70898cf-5c39-40b0-a6b8-0f49144c92b1","criterion":"reading-order-structure","criterion_name":"Reading Order & Structure","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"Section hierarchy and reading flow are preserved in the scanned paper, keeping headings and supporting paragraphs correctly connected despite the multi-column layout.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/a70898cf-5c39-40b0-a6b8-0f49144c92b1"},{"id":"527f7e79-f463-4b6e-b839-c5153a59a9f3","criterion":"table-preservation","criterion_name":"Table Preservation","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"The multicolumn table is reconstructed without losing its overall layout logic, so the table structure remains readable in the parsed output.","artifact_count":2,"evidence_url":"https://aidemos.com/evidence/527f7e79-f463-4b6e-b839-c5153a59a9f3"},{"id":"de43911e-dc72-4aee-a02a-2145d2d7acfb","criterion":"table-preservation","criterion_name":"Table Preservation","rank_role":"decisive","verdict":"failed","score":null,"score_total":null,"note":"Broken column boundaries and disrupted value alignment make the reconstructed table significantly less faithful to the source.","artifact_count":4,"evidence_url":"https://aidemos.com/evidence/de43911e-dc72-4aee-a02a-2145d2d7acfb"},{"id":"932e3296-1f2a-4d58-9bb4-5c7bb1e269a7","criterion":"text-ocr-completeness","criterion_name":"Text & OCR Completeness","rank_role":"decisive","verdict":"mixed","score":null,"score_total":null,"note":"The report says the parser recovers much of the underlying text from the scanned paper, but it does not present a measured completeness rate and the first-page hierarchy is still lossy.","artifact_count":3,"evidence_url":"https://aidemos.com/evidence/932e3296-1f2a-4d58-9bb4-5c7bb1e269a7"},{"id":"0f3a42b1-0d18-4374-bbeb-cd6776d7b79e","criterion":"visual-content-retention","criterion_name":"Visual Content Retention","rank_role":"decisive","verdict":"worked","score":null,"score_total":null,"note":"Charts are exposed through both page-wise markdown files and the extracted visual assets, keeping the visual content linked to its original document location.","artifact_count":3,"evidence_url":"https://aidemos.com/evidence/0f3a42b1-0d18-4374-bbeb-cd6776d7b79e"}],"appears_in":[{"page_type":"ranking","slug":"pdf-to-markdown-apis","title":"Best AI Tools to Convert Complex PDFs into Clean Markdown with an API","url":"https://aidemos.com/best/pdf-to-markdown-apis","binding":"run"}],"same_scenario":[{"id":"32fab15d-0885-4287-94a3-67e0d1d00611","tool":"adobe-api","tool_name":"Adobe API","verdict":"struggled","score":null,"score_total":null,"note":"Requires splitting an oversized scanned paper into two PDFs before processing, because the web interface rejects uploads over 1 MB."},{"id":"c8ea9121-f878-4e78-bf13-3ba3b4b6f0a5","tool":"extend-ai","tool_name":"Extend AI","verdict":"worked","score":null,"score_total":null,"note":"Handles the scanned research paper end-to-end, including multi-column prose, tables, charts, and handwritten marginalia, while still producing parsed markdown."},{"id":"9f0d06be-7ecb-4732-933c-ea6d52a1cc16","tool":"llamaparse","tool_name":"LlamaParse","verdict":"worked","score":null,"score_total":null,"note":"Processes a 12-page scanned paper end-to-end and reaches SUCCESS after extracting the page content and figure/table outputs."},{"id":"b50f09d7-7cee-4268-8441-50b46ddeb000","tool":"nutrient-io","tool_name":"Nutrient.io","verdict":"worked","score":null,"score_total":null,"note":"Processes an image-only scanned research paper end to end and returns a parsed markdown output plus preview, showing it can handle a multi-page scanned document."},{"id":"3bec7ed4-9b8e-44f3-b548-b76f0215b34c","tool":"reducto","tool_name":"Reducto","verdict":"struggled","score":null,"score_total":null,"note":"Handles the 12-page scan in 10.6 seconds with no truncation, but the densest 17-column table has unreadable stretches with injected glyphs and run-on numbers."},{"id":"ea443095-bb7a-45ec-92ca-6ff0afefe8e4","tool":"tensorlake","tool_name":"Tensorlake","verdict":"mixed","score":null,"score_total":null,"note":"On the scanned research paper, section flow and chart extraction work, but hierarchical tables degrade, so mixed-content handling is uneven rather than consistently robust."}]}