To read PDFs with Kimi, open Kimi, attach your PDF in a new chat, and ask for a summary or a specific explanation. Then open the relevant passage in the original file to check the answer. You can continue the conversation to explore sections, clarify terms, or compare information.
Last checked: October 12, 2026.
How to upload a PDF to Kimi
The steps below focus on the web version. Kimi’s getting-started guide supports uploading documents directly in chat. Open your PDF on your device first so you can confirm the file and keep it available for reference.
Before uploading a private document: Check that you have permission to share it. Kimi’s international privacy policy includes uploaded files in content used for service improvement and model training. Its terms explain how to request a training opt-out for future use. Review the conditions applicable to your account before sharing confidential material.
- Open Kimi in your browser, sign in if prompted, and start a New chat.
- Use the chat’s file-attachment control to select the PDF from your device.
- Wait for the upload to finish and check that the intended filename is attached.
- Send a clear request, such as the example below. Start with one document while you become familiar with the workflow.
Prompt to try: Use only the attached PDF. Explain its purpose, the main points I should understand, and any limitations the document itself mentions. Write for someone new to the topic. Include a relevant section heading or a page reference you can identify for each main point. Tell me which parts you could not read.
You can change the requested length or focus: ask for a short briefing, an explanation for a beginner, or the findings relevant to a particular decision. Kimi’s prompt guide recommends specifying the task and output format, then using follow-up questions to refine the response.
Ask follow-up questions and check page references
A summary helps you choose what to read next. Use a focused follow-up when you need to understand a claim, find an exception, or explain a difficult section. Mention the section title or quote a short phrase from the passage so Kimi has a clear target.
Prompt to try: Explain the main conclusion in plain English, using only this PDF. Show a brief passage that supports your explanation and identify its section. Include the PDF page position and printed page label if you can distinguish them. If the conclusion is not stated in the accessible content, say so instead of filling the gap.
Page references generated in an answer need checking. The page position shown by your PDF viewer can differ from the number printed on the page, especially when a report begins with a cover or introductory material.
- Open the referenced page and find the supporting passage.
- Read the surrounding paragraph, including any condition, exception, or footnote that changes its meaning.
- Compare important numbers, dates, names, and units with the original before using them.
If the page reference does not match, ask for the section heading and a short phrase you can search for in the PDF. If Kimi cannot identify the location, keep that uncertainty visible in your notes.
Also distinguish evidence from the PDF from outside information. A link to a website does not establish what your uploaded document says. For a document-only task, keep that restriction in your request and check the supporting passage yourself.
If Kimi reports that it could not read some pages, “not found” applies to the material it accessed. Check the relevant unreadable pages before concluding that the information is absent from the whole document.
Kimi PDF limits and free access
File size, file count, and document length
Kimi’s capability FAQ lists a maximum of 100 MB per file and up to 50 files at a time. Its overview describes the count as 50 files per session. Follow the limit displayed in your account, particularly when attaching a large batch.
These upload limits do not establish a fixed PDF page allowance. The amount of material Kimi can work with also depends on the model and conversation context—the information available for the current task. A file being accepted does not prove that every page has been fully examined.
For an ordinary reading task, begin with the model available in your account. If you need to work with a much longer document, check the applicable context access before switching models or paying for an upgrade.
Can you read PDFs with Kimi for free?
Kimi’s K2.6 page advertises free use, while its current credit rules say K2.6 chat also consumes membership credits. The pages do not establish a universal number of PDFs that every free account can read.
Check the allowance and model access shown in your account. The international membership guide documents avatar → Settings → Subscription on the web for checking your balance and refresh information. File-size limits and a remaining usage allowance are separate things.
Read scanned PDFs, tables, and charts
Try selecting a sentence in your PDF viewer. A selectable-text PDF gives you text to inspect and copy. An image-only scan contains pictures of pages; a mixed PDF may contain both. Check the specific page that matters to your question.
Kimi’s current chat documentation describes K3’s ability to understand text and visual content in PDFs and other office documents. For a scan, try the relevant page and compare a short passage with the original before relying on a larger extraction.
If reading fails, use a clearer, correctly oriented copy or provide the relevant pages separately. A version prepared with optical character recognition (OCR) can help when you need recognized text from a scan. Check that the preparation preserved the numbers and characters you need.
When a table or chart looks wrong
Ask Kimi to inspect the particular table or chart. For a table, the row labels, column headings, units, and footnotes are as important as the numbers. For a chart, check its axes, legend, and whether a value is printed explicitly or estimated from the graphic.
Correction prompt: Recheck the table you just used. Show its title, row labels, column headings, units, and relevant footnotes alongside the extracted values. Mark any unclear cell as unreadable. Keep blank cells separate from zero, and do not calculate totals until the values have been checked.
Compare the cells relevant to your question with the rendered page. If you need an exact figure from a chart, look for its underlying data table or a printed label before accepting an estimate.
Handle long PDFs or compare two documents
Read a long document in useful sections
For a lengthy report, ask for an overview of its structure, then work through the sections relevant to your goal. Give each request a clear boundary, such as a named chapter or the pages containing the result you want to understand.
If a file exceeds the available context, Kimi’s troubleshooting guide recommends splitting it. Split at meaningful chapter or section boundaries and retain the original filename and page range in your notes. This helps you trace an answer back to the complete PDF.
When a conversation itself becomes too long, save the useful conclusions and source locations, then continue in a new chat with that summary and the relevant file. Check any carried-over claim against its source if it affects the next task.
For ongoing work, a Kimi Project can keep reference files and related chats together. Project files are read on demand, so continue naming the document and section you want Kimi to use.
Keep evidence from two PDFs separate
When comparing versions, check the filenames, dates, and editions before uploading. Define what matters: changed instructions, different findings, revised figures, or added conditions. Ask Kimi to attribute each finding to the correct document.
Comparison prompt: Compare the two attached PDFs for changes in their instructions and recommendations. Use each filename as a column heading. For every difference, identify the relevant section in each file and give a page reference where available. Distinguish an explicit change from information you could not find or read.
Review the cited sections side by side. If only one PDF contains a statement, establish whether it was removed, moved elsewhere, or simply missed before describing it as a substantive change.
If you will reuse the same PDFs across several related chats, see how Kimi Projects organizes shared files and project instructions; still verify which passages support each answer.
What to do when Kimi cannot read your PDF
Start with the visible problem and the error message, if one appears. The following checks help narrow down what to try next; they do not assume every failure has the same cause.
| Problem | What to try next |
|---|---|
| The PDF is rejected | Confirm that it opens locally and check its size. Try one file in a new chat. If it is encrypted, obtain a readable copy you are authorized to use. |
| The upload does not finish | Check the connection, refresh the page if necessary, and try the file again. Record any error message rather than repeatedly resubmitting the same request. |
| A scan or page is unreadable | Inspect its orientation and clarity. Try a clearer copy, the relevant pages, or an OCR-prepared version, then verify the recognized text. |
| The answer misses an important section | Name the section or provide its pages directly. Ask Kimi to state what it could access before requesting another conclusion. |
| A number or quotation appears wrong | Ask for the exact supporting passage or table location. Compare it with the original and request a correction tied to that evidence. |
| An access or credit limit appears | Check your account’s remaining allowance and the stated refresh or access conditions. A plan change will not repair a damaged PDF or an unclear scan. |
If the problem continues, note the file size, the model you selected, the interface you used, and the exact error text. Share a non-sensitive example with support if needed, rather than sending a confidential document as a troubleshooting attachment.
If a small, readable PDF still fails, use the Kimi troubleshooting guide to check sign-in, upload, browser, and app problems.
Other ways to work with your PDF
Use a publicly accessible PDF link
Kimi’s Search guide supports fetching public PDF URLs. Paste the link into the conversation and specify the question or summary you want. Ask Kimi to identify the document it accessed so you can check that it reached the intended file.
If the link requires a login or cannot be fetched, download an authorized copy and attach it. A path to a file on your computer does not provide Kimi with the file itself.
Save a formatted summary
You can keep useful answers in your own notes, including the source filename and relevant page or section. If you need a formatted document, Kimi Docs provides a workflow for creating downloadable outputs from source material.
Use our Kimi Docs guide when you need a formatted downloadable document.
Ask for the format you need, such as a Word or PDF summary, and specify the sections and source references to retain. Open the downloaded file to check its contents and layout. Keep the original PDF separately so you can return to the source when reviewing the summary.