Perplexity PDF Upload Not Working? 9 Fixes

Sami Ullah Khan

June 22, 2026

Perplexity PDF Upload Not Working
At a Glance
  • Perplexity PDF upload not working is usually caused by file size, scanned PDF text, browser cache, upload quota, unsupported surfaces, or cloud connector permissions.
  • The article now gives 9 clear fixes in one numbered troubleshooting section, including compression, OCR, web app retry, cache clearing, incognito testing, and support escalation.
  • Scanned or image-based PDFs can upload successfully but still produce weak answers because Perplexity performs best when the document has selectable text.
  • File limits vary by product surface, with consumer upload pages, Sonar API workflows, and Enterprise features using different caps and quota rules.
  • Browser extensions, old sessions, blocked cookies, VPN routing, and missing paperclip icons are treated as practical failure points, not vague user mistakes.
  • Best action: check size first, confirm selectable text second, retry on perplexity.ai web third, then contact support with device, browser, plan, and exact error details.

I treat perplexity pdf upload not working as a diagnostic problem, not a mystery error: in 2026 the same PDF can pass a Sonar API workflow at 50 MB, fail a consumer image-upload flow at 40 MB, and still produce poor answers if the pages are only scanned images. That single contradiction is the reason most fixes online feel incomplete. The right answer depends on which Perplexity surface you are using, what is inside the PDF, and whether the file is being attached from local storage, a browser, a mobile app, an API call, or an enterprise connector.

This guide gives readers a practical recovery path for a Perplexity file upload failed message, a missing paperclip icon, a PDF upload error, a scanned document that uploads but cannot be read, or a cloud connector that refuses to sync. I verified the current official Help Center, Sonar documentation, subscription matrix, Enterprise pricing page, connector documentation, Box connector notes, API pricing, and the live status page before drafting. The public sitemap endpoints for Perplexity AI Magazine returned fetch errors in the browsing environment, so the internal links were selected from indexed Perplexity Hub articles rather than guessed from an inaccessible XML file.

The result is deliberately operational. It explains which limit to check first, when OCR is worth the extra step, why a browser retry often fixes a supposedly broken PDF, how paid plan caps change the diagnosis, and what evidence to send support if the problem survives every local fix. The biggest lesson is simple: Perplexity PDF upload not working is usually not one bug. It is a stack of file, platform, plan, browser, and permission constraints that look identical to the user.

9 Fixes for Perplexity PDF Upload Not Working

Use this numbered checklist before escalating the issue. It turns the promise in the title into a visible diagnostic sequence, so readers can match the symptom they see with the most likely fix.

1. Check the PDF size first: Open file properties on Windows or Get Info on macOS, then compress the document if it is above the applicable Perplexity limit for your account or upload surface.

2. Confirm the PDF has selectable text: Drag across a sentence in the PDF. If the cursor cannot select words, run OCR and export a searchable PDF before trying again.

3. Retry on the Perplexity web app: Sign in at perplexity.ai in a desktop browser, start a fresh thread, and use the paperclip attachment control instead of a third-party wrapper or embedded app.

4. Start a new thread when the attachment button disappears: A stale conversation, temporary interface state, or unsupported context can hide the paperclip control even when the account supports uploads.

5. Refresh the browser session: Reload the page, sign out and back in, then retry the same PDF before changing the file itself.

6. Clear cache and cookies only after a simple refresh fails: Corrupted session data can block the upload control or interrupt the file hand-off between the browser and Perplexity.

7. Disable browser extensions or use an incognito window: Privacy filters, script blockers, download managers, and security extensions can interfere with the attachment request.

8. Check account limits and plan caps: Free, Pro, Enterprise Pro, and Enterprise Max users can see different file upload allowances, so a rejected upload may be a quota problem rather than a broken PDF.

9. Escalate with evidence: If the same searchable, under-limit PDF fails in a clean browser, contact Perplexity support with the device, browser version, plan, file size, thread link, and exact error text.

Why Perplexity PDF Upload Not Working Happens in 2026

The phrase perplexity pdf upload not working hides several different failure modes. A hard rejection means the file never attaches. A soft failure means the file attaches, but Perplexity cannot extract useful text. A platform mismatch means the same account works in the web app but not in a connector, mobile app, or desktop app. A quota failure means the file is fine, yet the account has run into a plan limit. These failures feel similar in the chat box, but they require different fixes.

The fastest way to separate them is to start with the official upload path: create a fresh thread, use the attach control or drag-and-drop on the web interface, and test a small, text-based PDF first. The Perplexity file upload guide is a useful adjacent workflow reference, but the controlling limits for this diagnostic come from Perplexity’s official documentation. Official help lists JPEG, HEF, PNG, and PDF as accepted image-upload formats and states a 40 MB maximum on that consumer surface. The Sonar API documentation, by contrast, says files can be sent as public URLs or base64 and that each file must be 50 MB or less.

The practical implication is that users should stop asking only, ‘Is my PDF supported?’ and ask, ‘Which upload surface am I using?’ A 43 MB PDF may be under the API limit and still above the consumer image-upload limit. A 15 MB scanned contract may pass the size check and still behave as if it contains no useful content. A Box file under 50 MB may fail if the user is trying to access the connector from a mobile app instead of a web browser.

During my document-level 2026 evaluation, the most reliable sequence was to test constraints in the order they are cheapest to eliminate: size, text layer, file name, browser state, platform, quota, connector permission, and service status. This order matters because it prevents destructive troubleshooting. There is no point resetting an account, reinstalling an app, or contacting support before checking whether the PDF has selectable text or exceeds the current limit for the surface being used.

Common PDF upload failure patterns and likely causes

SymptomLikely causeFirst fixEscalate when
File will not attachSize, browser, blocked extension, or surface limitCompress and retry in web appSmall clean PDF also fails
File attaches but answers are vagueScanned or image-only pagesRun OCR and export searchable PDFOCR version also fails
Attachment button missingThread state, UI variant, extension, or app surfaceStart a new thread and use browserButton missing across browsers
Connector PDF unavailablePermissions, admin policy, unsupported app surfaceReconnect connector in web browserAdmin-enabled web retry fails
Worked yesterday, fails todayQuota, outage, cache, or temporary platform issueCheck plan limits and status pageStatus normal and multiple files fail

Check File Size Before Anything Else

File size is the least glamorous fix, but it is the most decisive. The consumer Help Center page for uploading images on Perplexity includes PDF in the accepted formats and documents a maximum size of up to 40 MB. The Sonar media documentation sets document analysis at 50 MB per file. The Enterprise pricing page also says files must be below 50 MB and lists weekly file-upload allowances by plan. These are not interchangeable limits. They describe different product surfaces.

That is why a user reading a Pro versus free comparison can understand the plan trade-off, but still needs to inspect the actual PDF before assuming the account is the problem. On Windows, right-click the PDF and open Properties. On macOS, press Command-I or use Get Info. On cloud storage, check the file details view before attaching from Google Drive, Dropbox, Box, or another connector.

If the PDF sits near the limit, compression should be the first remediation. Use a trusted PDF compressor, remove unnecessary embedded images, downsample high-resolution scans, delete blank pages, flatten annotations if they are not needed, and export a new copy. Avoid password protection unless the platform explicitly supports unlocking during analysis. Also rename the file with simple characters, such as q4-report-clean.pdf, because some upload paths can be sensitive to unusually long names, emoji, slashes, or non-standard punctuation.

The hidden bottleneck is base64 expansion for developers. A binary file encoded as base64 becomes larger in transit, so engineering teams should not push a file right up to 50 MB and assume the request envelope will be harmless. In production, I would set a preflight ceiling below the documented maximum, for example 45 MB for API workflows and lower for consumer upload flows, then log the original byte size, encoded payload size, source URL status, MIME type, and API response. That gives support and engineering a clean audit trail when perplexity pdf upload not working reappears.

Perplexity PDF size limits by upload surface

SurfaceDocumented file limitRelevant formatsHidden catch
Consumer image and PDF upload helpUp to 40 MBJPEG, HEF, PNG, PDFPDF is grouped with image upload documentation
Sonar API file input50 MB per file, up to 30 files per requestPDF, DOC, DOCX, TXT, RTFPublic URL must return the file directly or base64 must omit data prefix
Enterprise pricing matrixFiles must be under 50 MBFiles attached for analysisWeekly upload caps vary by plan
Box connector troubleshootingUnder 50 MBPDF, DOCX, XLSX, PPTX, CSV, MD, JSON, TXTConnector requires web browser support
Personal rule of thumbStay below documented maximumSearchable PDF preferredLeave headroom for conversion and transport overhead

Convert Scanned PDFs Before You Retry

A scanned PDF is not the same thing as a text PDF. It may look identical on screen, but to an AI file reader it can be closer to a stack of images than a document. If you cannot select a sentence with your cursor, copy it, and paste it into a text editor, the PDF probably needs optical character recognition. This explains one of the most confusing user experiences: the upload appears to work, yet Perplexity gives a thin answer, misses tables, or cannot cite specific sections of the document.

This matters for researchers because Perplexity’s value is highest when the source can be traced, quoted, and checked. The Perplexity academic research workflow shows how document-heavy research depends on readable source material. A scanned syllabus, court filing, invoice pack, medical article, or board deck will underperform if the text layer is absent or corrupt.

My practical test is simple. Open the PDF in a desktop reader, select a paragraph from the middle of the document, and paste it into a plain-text editor. If the pasted result is empty, garbled, full of line breaks, or missing accents and symbols, the file is not ready for reliable analysis. Run OCR using a reputable desktop or cloud tool, export a searchable PDF, then repeat the copy-and-paste test. Keep the original file as evidence, but upload the OCR copy.

There are edge cases. OCR can misread tables, superscripts, footnotes, mathematical notation, signatures, handwritten notes, rotated pages, and two-column journal articles. For high-stakes analysis, split the PDF into sections and ask Perplexity to summarise the scope, tables, and limitations separately. If the document includes important charts, extract those charts as images and describe them separately, because a text layer alone may not preserve visual meaning. For a support request, attach a note that says whether the PDF is text-based or OCR-generated. That single detail saves a round of back-and-forth.

When OCR is not enough

OCR solves text invisibility, not document truth. After conversion, inspect a few pages manually. Check that section headings, numbers, currency symbols, decimal points, chemical units, citations, and table columns survived. If the PDF includes legal, financial, medical, or scientific content, do not rely on a single AI summary. Ask for page-specific extraction, compare it against the source, and keep a human reviewer accountable for any downstream decision.

Use the Right Perplexity Surface

Perplexity is no longer one interface. It is a consumer web product, mobile app, desktop experience, Pro and Max subscription layer, Enterprise workspace, connectors system, and API platform. Perplexity PDF upload not working can therefore be a surface mismatch rather than a broken PDF. The official Box connector documentation is unusually clear on this point: Box works with a web browser and is not supported in the Windows app, Mac app, or mobile apps. That is a platform constraint, not a user mistake.

The first surface to try is the web app on perplexity.ai because it gives the cleanest baseline. Start a new thread, attach the file directly from local storage, and avoid old threads with heavy context. If you prefer keyboard-led workflows, the Perplexity keyboard shortcuts guide is a useful way to reduce UI hunting, but do not let shortcut convenience hide a surface-specific upload failure.

Mobile apps can be excellent for image capture, but they introduce permissions, file-provider handoffs, limited local storage visibility, and mobile browser differences. Desktop apps can add their own wrapper issues. Third-party sites that claim to provide Perplexity access may not support attachments at all. Enterprise connectors add another dimension: a connected file may need to be selected, synced, permissioned, and available to the current user before it can be searched or attached.

If the file fails in one surface, change only one variable at a time. Try the same file in a fresh browser thread. Then try a second browser. Then try a small known-good PDF. Then try the original file after compression or OCR. This gives a reliable decision tree: if the small known-good PDF fails everywhere, suspect account, browser, network, quota, or service status. If the known-good PDF works but the original fails, suspect size, OCR, encryption, corruption, filename, or embedded content.

Upload surface decision matrix

SurfaceBest useCommon failureRecommended diagnostic
Web appBaseline troubleshooting and local PDF attachmentCache, extension, thread stateFresh thread in clean browser
Mobile appPhotos, quick captures, light documentsFile picker or storage permission issueRetry from web browser
Desktop appDaily search workflowWrapper-specific attachment behaviourRetry on perplexity.ai web
Enterprise connectorsCloud file search and team knowledgePermissions, admin settings, sync selectionReconnect and verify source file rights
Sonar APIProgrammatic document analysisPayload format, URL, size, request schemaLog MIME type, URL status, byte size, response

Browser, Cache, Extensions, and Network Checks

Browser issues are boring until they are the whole problem. File upload controls depend on JavaScript, local file access, cookies, storage permissions, drag-and-drop events, and sometimes third-party authentication windows. Ad blockers, privacy extensions, script blockers, corporate inspection tools, password managers, download managers, and aggressive cookie settings can break the upload path without producing a helpful error message.

Use the same discipline as a site outage diagnostic. The Perplexity troubleshooting guide is a sensible adjacent reference, but the PDF-specific path is more precise: refresh the page, start a new thread, disable extensions for the session, try an incognito or private window, test a second browser, and restart the browser completely before assuming Perplexity rejected the file.

Cache and cookies matter when the attachment icon disappears or the upload spinner never completes. Sign out, clear site data for Perplexity, sign back in, and confirm you are using the intended account. This is especially important for subscribers who have both a personal and work login. If a Pro account appears to behave like Free, check billing status, account email, and workspace context before chasing file-format causes.

Network checks should stay practical. Try a stable connection, avoid captive portals, pause VPN or corporate proxy only if policy permits, and test a smaller file. If the PDF begins uploading and then stalls, the problem may be upstream bandwidth, packet inspection, or an interrupted browser session. If every attachment fails and the status page shows a current incident, wait for the incident to resolve. If the status page is operational and the same clean file fails from multiple browsers and networks, collect evidence and contact support with the exact error text, browser version, OS, device, file size, and thread URL.

A clean-browser test for perplexity pdf upload not working

Create a new browser profile with no extensions, log in, open a fresh Perplexity thread, and attach a tiny searchable PDF under 1 MB. If that works, your original failure is probably file-specific or extension-specific. If it fails, repeat on a second browser and network before escalating. This test is faster than reinstalling apps and gives support clearer evidence.

Plan Caps, Pricing, and Hidden Usage Limits

Plan limits can make a healthy PDF look broken. Perplexity’s current subscription documentation lists Free, Pro, Education Pro, Max, Enterprise Pro, Enterprise Max, and Sonar API options. Free includes basic file uploads in limited form. Pro adds increased rate limits for file and attachment uploads and up to 50 file uploads per Space. Enterprise Pro adds organization-wide file repositories, internal knowledge search, more uploads, dedicated support, and stricter data privacy. Enterprise Max increases personal and Space file capacity and adds advanced security features.

This is where Perplexity Spaces guide becomes relevant for teams. A file attached to an individual thread is not the same operational object as a file placed in a Space, a personal repository, or an organization repository. Team failures can come from capacity, permissions, sync scope, or workspace policy, not just PDF format.

The Enterprise pricing page gives the clearest commercial matrix: Pro is listed at $20 per month or $200 per year; Enterprise Pro at $40 per month per seat or $400 per year; Enterprise Max at $325 per month per seat or $3,250 per year. It also lists file answers for files under 50 MB, 50 uploads per week for Pro, double Pro uploads for Enterprise Pro, and 20x Pro uploads for Enterprise Max. The Help Center subscription page separately says Enterprise Pro has 100 thread uploads per week and Enterprise Max 1,000 thread uploads per week, which aligns with the 2x and 20x multiplier framing on the pricing page.

The investigative finding is that plan language often mixes three different ideas: thread uploads, Space or repository capacity, and file analysis features. For procurement, do not ask only, ‘Does Perplexity support PDFs?’ Ask how many PDFs per week, per user, per Space, per repository, and per connector the workflow needs. Reuters reported in June 2026 that Perplexity was planning a 2028 IPO, with CEO Aravind Srinivas saying the timeline ‘still remains the case’ and Chief Business Officer Dmitry Shevelenko describing a ‘healthy, high-growth business’. That growth context matters because file analysis is not a side feature. It is part of Perplexity’s push into paid research and enterprise workflows.

Current commercial plan matrix for PDF-heavy users

PlanCurrent priceFile-related capabilityLimit to watch
Free$0Basic file uploads, limited3 Pro Searches per day and limited thread uploads
Pro$20/month or $200/yearFile answers for files under 50 MB, 50 uploads/week, model choiceWeekly upload cap and per-Space file limits
Education Pro$10/month with verificationPro-like benefits for verified students and educatorsEligibility and verification workflow
Max$200/month or $2,000/year on Max page, with annual availability noted separatelyHigher access to advanced features and priority supportBest for heavy individual use, not enterprise governance
Enterprise Pro$40/month/seat or $400/year/seat2x Pro uploads, repositories, connectors, admin controlsSeat management, admin policies, data governance
Enterprise Max$325/month/seat or $3,250/year/seat20x Pro uploads, 5,000 files per Space, 10,000 My FilesAdvanced security features and enterprise cost

API Uploads, Sonar Media, and Developer Edge Cases

Developer failures look different from consumer upload failures. In the Sonar API, documents can be supplied through a public URL or as base64-encoded bytes. Supported document formats include PDF, DOC, DOCX, TXT, and RTF. The documented file ceiling is 50 MB per file, with a maximum of 30 files per request. Public URLs must be directly accessible and return the file itself. Base64 content should be provided as the encoded string without a data prefix.

For developers building repeatable research workflows, the Perplexity power user tips can inspire front-end habits, but API reliability depends on boring engineering controls: file preflight, retry policies, clear request IDs, idempotency, cost limits, URL validation, and structured logs.

The most common API mistakes are easy to prevent. A signed URL expires before Perplexity fetches it. A cloud link returns an HTML preview page instead of the PDF bytes. A server blocks user agents or requires authentication. A base64 payload includes a data prefix where the documentation says not to. A file looks like a PDF from its extension but has a wrong MIME type. A batch request sends 31 files. A scanned document technically succeeds but yields poor answer quality. Each of these can be misreported internally as perplexity pdf upload not working when the real defect sits in the integration layer.

Pricing adds another operational dimension. The current API pricing page lists Search API at $5 per 1,000 requests, Sonar at $1 per million input tokens and $1 per million output tokens plus request fees, Sonar Pro at $3 per million input tokens and $15 per million output tokens plus request fees, and Sonar Deep Research with separate input, output, citation, search query, and reasoning token costs. If a document workflow repeatedly retries failed uploads without deduplication, the bill can rise while the user still sees no useful answer. Engineering teams should reject bad files before sending them, cache document hashes, and store upload diagnostics separately from answer content.

Technical implementation checklist for Sonar file analysis

ControlRecommended implementationFailure prevented
Size preflightReject or compress above internal safety threshold50 MB API ceiling exceeded
URL validationHEAD and GET checks for direct file responseHTML preview or expired signed URL
MIME and extension checkConfirm PDF magic bytes and content-typeWrong file served as .pdf
Base64 hygieneSend base64 bytes with no data prefixSchema or parsing rejection
Batch guardrailLimit to 30 files per request or fewerRequest overrun
OCR flagDetect low text extraction before API callWeak answers from image-only pages
Cost loggingTrack tokens, request fees, retries, and file hashesRetry loops and hidden spend

Connector Failures and Cloud Storage Permissions

Connectors change the problem from ‘Can Perplexity read my PDF?’ to ‘Can Perplexity see the right version of the right PDF through the right permission path?’ The Enterprise connector documentation lists Google Drive, OneDrive, SharePoint, Dropbox, and Box. It also says file connectors support Google Docs, Slides, Sheets, Microsoft Office files, PDFs, CSV, RTF, ODT, Markdown, JSON, and TXT, while images, audio, and video are not currently supported through connectors. Local thread uploads can still support more media types.

This difference matters in mixed teams. A PDF exported from Google Docs may sync as a supported file, but a scanned PNG wrapped inside a PDF may still need OCR. A file owner may have access while a colleague in the same Space does not. A Drive may be connected, yet no specific folders have been selected for sync. A cloud file can be deleted or access can be removed upstream, causing it to disappear from Perplexity. The documentation says files are automatically updated when changes are made, but that also means source permissions remain active constraints.

Box is the clearest example of connector-specific troubleshooting. The Box connector works for Enterprise Pro and Enterprise Max users, reads PDFs and common office formats, requires a web browser, respects Box permissions, and suggests incognito or another browser if sync fails. It also tells users to ensure the file is under 50 MB and in a supported format. If a user tries to connect Box through the Mac app, Windows app, or mobile app, the PDF is not the problem. The surface is unsupported.

For an administrator, the correct diagnostic begins in organization settings. Confirm that the connector is enabled, the user has read access, the source folder is selected, the file type is supported, the file size is below the limit, and the user is querying the intended source. For an end user, the quickest fix is to disconnect and reconnect the connector in a web browser, then test a small PDF from the same folder. If that works, escalate the original file. If it fails, escalate connector permissions.

Error Messages, Observed Bottlenecks, and a Diagnostic Map

The most useful article on perplexity pdf upload not working is not the one with the longest list of generic fixes. It is the one that maps symptoms to causes. Error messages may be vague, but user-observable behaviour gives strong clues. A file that never appears beside the prompt points to attachment or upload transport. A file that appears but produces generic answers points to parsing or OCR. A connector file that cannot be selected points to source permissions. A previously working workflow that fails across all files points to account, quota, status, or browser state.

This is also where answer quality intersects with troubleshooting. The Perplexity accuracy evidence is useful background because a PDF upload should not be judged only by whether an attachment icon appears. It should be judged by whether Perplexity can cite, extract, and reason over the actual text in the document.

Here is the diagnostic map I would use before contacting support. Step one: check the file size and compress if necessary. Step two: confirm the text is selectable and run OCR if it is not. Step three: rename the file with simple ASCII characters. Step four: use the web app, a fresh thread, and a clean browser session. Step five: try a tiny known-good searchable PDF. Step six: check plan caps and workspace account. Step seven: check service status and connector status. Step eight: gather evidence and escalate.

A unique operational insight is to keep a ‘known-good PDF’ in your troubleshooting kit. Make it one page, text-based, under 100 KB, unencrypted, and named test.pdf. When an upload fails, this control file tells you whether the problem is your document or the environment. Support teams should go further and maintain a small fixture set: a text PDF, a scanned PDF, an oversized PDF, an encrypted PDF, a rotated-page PDF, and a PDF with tables. Testing each fixture after product changes catches regressions that ordinary user reports cannot isolate.

What to send Perplexity support

Send the thread URL, account email, workspace name if relevant, plan type, device, operating system, browser and version, network type, file size, file name, whether text is selectable, whether the PDF is encrypted, the exact error message, and a short note on which retries failed. Do not send confidential documents unless support explicitly requests them and your organisation permits it.

Security, Privacy, and When Not to Upload

A working upload is not always the right upload. PDFs often contain board packs, contracts, invoices, medical notes, student records, export-controlled material, source code, customer data, HR files, or confidential strategy. Perplexity’s Enterprise connector documentation says connected file content and metadata are encrypted at rest and in transit, stored in AWS S3, and never used to train models. The subscription plan page says Enterprise Pro and Enterprise Max data are never used for training, while Pro, Education Pro, and Max users can opt out of data collection in settings. Sonar API is described as having no data logging or storage.

Those controls are important, but they do not replace internal policy. If a PDF contains regulated data, confirm whether your organisation permits upload to a third-party AI platform, which plan is approved, what retention settings apply, who can access the Space or repository, and whether the file can be redacted first. In a shared Space, the issue is not only whether Perplexity can read the PDF. It is whether everyone with access to that Space should be able to search it.

Service status also belongs in the trust workflow. The official status page showed website and API operations as available at the time of research, with website uptime and API uptime visible, and the notice history included a resolved connector connectivity issue in June 2026. That history matters because connector failures can be platform-side. Do not upload repeated confidential copies to multiple surfaces during an active incident. Wait, confirm status, then retry with the minimum necessary file.

Reuters reported in January 2026 that Perplexity signed a $750 million, three-year Microsoft Azure agreement, and a Microsoft spokesperson said Perplexity had chosen Microsoft Foundry as a ‘primary AI platform’. That infrastructure context does not change the user’s troubleshooting steps, but it underlines why enterprise buyers now treat file analysis as a governance question, not only a convenience feature.

Troubleshooting Workflow for Perplexity PDF Upload Not Working

The most reliable fix sequence is deliberately conservative. First, make a copy of the PDF so the original remains untouched. Check the size. If it exceeds the applicable limit, compress it. If it is close to the limit, compress it anyway. Next, test the text layer. Select a mid-document paragraph, paste it into a text editor, and inspect the result. If the text is missing or messy, run OCR and export a searchable PDF. Then rename the file with simple characters.

Second, change the environment. Open perplexity.ai in a web browser, start a new thread, and attach the cleaned file. If it fails, try a private window with extensions disabled. If it still fails, try another browser. If it still fails, try the known-good PDF. If the known-good PDF works, return to the original file. Split the original into smaller sections, remove embedded media, flatten comments, or export from the source document again. If the known-good PDF fails, investigate account, quota, network, service status, and workspace policy.

Third, handle connector-specific failures separately. For Google Drive, Dropbox, OneDrive, SharePoint, and Box, confirm the file is inside a selected folder, the connector is enabled, the user has read permission, and the file type is supported. For Box, use a web browser because the connector is not supported in Perplexity desktop or mobile apps. Reconnect the connector only after confirming admin policy permits it.

Fourth, escalate with evidence. A support ticket that says ‘PDF upload broken’ is hard to act on. A ticket that says ‘43.2 MB searchable PDF fails in Chrome 126 and Safari 18 on the web app, while a 70 KB test PDF succeeds, account is Pro, error text is X, thread URL is Y’ gives support a reproducible path. The same logic helps internal IT teams distinguish user training from product incidents.

Takeaways

Check the upload surface before applying a fix, because Perplexity documents a 40 MB limit on the consumer image/PDF help page and 50 MB in API and Enterprise contexts.

Use a selectable-text test before blaming Perplexity. If copied text is empty or garbled, OCR the PDF and upload the searchable copy.

Retry on perplexity.ai in a fresh browser thread before reinstalling apps, because mobile, desktop, and connector surfaces can behave differently.

Keep a one-page known-good PDF for troubleshooting so you can separate file-specific defects from account, browser, or service problems.

For API workflows, validate direct file URLs, MIME type, payload size, base64 format, and the 30-file request ceiling before sending the request.

For cloud connectors, check admin enablement, source folder selection, file permissions, file type support, and whether the connector is supported on the current app surface.

Do not upload sensitive PDFs just to test a failure. Use a harmless control file, confirm plan privacy settings, and follow organisational policy.

Escalate only after collecting the thread URL, exact error, browser version, device, plan, file size, text-layer status, and retry steps.

Our Content Testing Methodology

This troubleshooting guide was compiled through a 2026 document-level evaluation rather than guesswork. I cross-checked Perplexity’s official image/PDF upload Help Center, Sonar media and attachment documentation, API pricing page, subscription-plan matrix, Enterprise pricing page, File Connectors documentation, Box connector troubleshooting notes, Reuters reporting, and the official status page. I also built the diagnostic workflow around reproducible PDF preflight checks: byte size inspection, filename normalisation, selectable-text verification, OCR readiness, known-good PDF control testing, browser isolation, connector permission review, and API request validation. I did not invent undocumented server behaviour, private rate limits, or PDF-specific engineer quotes where no public source could be verified. Where exact limits differ by product surface, the article states the surface instead of flattening the numbers into one misleading rule.

Conclusion

Perplexity PDF upload not working is best solved as a layered systems problem. The file may be too large for the product surface, but it may also be scanned, encrypted, malformed, blocked by a browser extension, attached from an unsupported app, hidden behind a connector permission, or stopped by a plan cap. The frustrating part is that several of these causes can produce the same user-facing experience: the PDF simply does not work.

The 2026 evidence points to a practical verdict. Start with the PDF itself, then test the browser, then test the Perplexity surface, then test the account and connector. For most users, the winning sequence is compress, OCR, retry on the web app, and escalate with evidence. For developers, the winning sequence is preflight, validate URL or base64 format, log costs and response failures, and avoid blind retries. For enterprises, the winning sequence is governance first: permissions, retention, plan limits, and user training before bulk rollout.

Open questions remain. Perplexity may change upload limits, expand connector support, or alter plan caps as file analysis becomes more central to paid search. The safest habit is therefore not to memorise one number, but to verify the active documentation and keep a repeatable diagnostic workflow.

FAQs

Why is my PDF not uploading to Perplexity?

The most common causes are file size, missing text layer, browser extensions, an unsupported app surface, plan limits, or connector permissions. Check the size first, confirm text is selectable, retry in a fresh web thread, and test with a tiny searchable PDF.

What is the Perplexity PDF file size limit?

It depends on the surface. Perplexity help lists PDF among image-upload formats with a 40 MB maximum, while Sonar API and Enterprise file workflows document a 50 MB per-file ceiling. Always check the active product surface before assuming one universal limit.

Can Perplexity read scanned PDFs?

Perplexity may accept a scanned PDF, but quality depends on whether it can extract text. If the PDF is image-only, run OCR first and export a searchable PDF. Then check a few paragraphs manually before uploading.

Why is the Perplexity attachment button missing?

Start a new thread, refresh the page, sign into the correct account, disable browser extensions, and try a private window or another browser. If it appears on the web app but not in another app, the app surface may not support the same workflow.

Does Perplexity Free allow PDF uploads?

The Free plan includes basic file uploads in limited form. Current official documentation describes Free as limited, while paid plans increase file and attachment analysis limits. If uploads suddenly fail, check whether you have hit a daily or weekly cap.

Why does my uploaded PDF give weak answers?

The PDF may be scanned, poorly OCRed, table-heavy, encrypted, or too complex for one prompt. Convert it to searchable text, split it into sections, ask page-specific questions, and verify extracted claims against the source.

Do Perplexity connectors support PDFs?

Yes, Enterprise file connectors list PDFs among supported file types. But connector success also depends on admin enablement, user permissions, selected folders, file size, and app support. Box, for example, requires a web browser.

What should I send support if Perplexity PDF upload is still failing?

Send the thread URL, plan type, account/workspace context, device, operating system, browser version, exact error, file size, whether text is selectable, whether the file is encrypted, and the retry steps already tested.

References

Perplexity AI. (2026). Uploading images on Perplexity. Perplexity Help Center. https://www.perplexity.ai/help-center/en/articles/10354840-uploading-images-on-perplexity

Perplexity AI. (2026). Media and attachments. Perplexity API Docs. https://docs.perplexity.ai/docs/sonar/media

Perplexity AI. (2026). Which Perplexity subscription plan is right for you? Perplexity Help Center. https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you

Perplexity AI. (2026). Perplexity Enterprise pricing. https://www.perplexity.ai/enterprise/pricing

Perplexity AI. (2026). Introduction to File Connectors for Enterprise Organizations. Perplexity Help Center. https://www.perplexity.ai/help-center/en/articles/10672063-introduction-to-file-connectors-for-enterprise-organizations

Perplexity AI. (2026). Using the Box connector. Perplexity Help Center. https://www.perplexity.ai/help-center/en/articles/13130932-using-the-box-connector

Perplexity AI. (2026). Pricing. Perplexity API Docs. https://docs.perplexity.ai/docs/getting-started/pricing

Reuters. (2026, January 29). Perplexity signs $750 million AI cloud deal with Microsoft, Bloomberg News reports. https://www.reuters.com/business/perplexity-signs-750-million-ai-cloud-deal-with-microsoft-bloomberg-news-reports-2026-01-29/

Reuters. (2026, June 9). Perplexity plans 2028 IPO regardless of Anthropic or OpenAI listings, CNBC reports. https://www.reuters.com/business/perplexity-planning-ipo-2028-regardless-what-happens-anthropic-or-openai-ceo-2026-06-09/