# Compress PDF to X MB — yeezhan.com > One-time pay-per-task emergency tool to make your file ready for submission. > No credits, no balance. One payment unlocks one task result, valid for 24 hours. Tool domain: https://compress-pdf.yeezhan.com Skill definition: https://compress-pdf.yeezhan.com/skill/SKILL.md API documentation: https://compress-pdf.yeezhan.com/api/docs ## What this tool does Shrink a PDF to the exact size cap an upload portal demands — verified before you pay a cent. ## Typical user problems - Upload portals hard-fail the moment your PDF is one kilobyte over the limit — job applications, university submissions, visa portals, RFP systems. You do not need a PDF course, you need the file under the cap in the next two minutes, and you need to know it will pass. - Government portals, recruitment systems and university application sites love the 1 MB hard cap. A single scan blows past it. You need the file at 0.9x MB, verified, in one attempt. - 2 MB is the most common upload ceiling across application forms and email gateways. You need the file under it without destroying readability. - Visa portals and tender submission systems commonly reject anything over 5 MB. You need the whole document — every page — under the cap. - Email attachment limits sit at 10 MB total. Your PDF must land under 9-ish MB with margin for the rest of the message, or the mail bounces. ## Parameters - target_mb: number (range 0.05-500MB), required. Target size. The hard cap the upload portal enforces - allow_rasterize: boolean, optional. Allow image-mode fallback. If the cap is unreachable with text preserved, allow rebuilding pages as high-quality images (text becomes non-selectable) ## Pricing - Watermarked preview: free, no login required - Final clean output: $0.99 one-time per task - Require login (email/phone OTP) to complete payment - Result available for download 24 hours after payment ## Limitations - Maximum input size: 200 MB - Encrypted / password-protected PDFs are not supported - Target sizes below ~50 MB: practical floor depends on page count and image content - Image-mode fallback makes text non-selectable (the result page always discloses the strategy used)