Receipts, screenshots, whiteboards, handwritten notes, foreign menus, scanned PDFs.
Drop one in — get clean, copyable text in 2 seconds. No app. Free sign-up with Google in one click.
Drop an image here, or click to upload
Supports JPEG, PNG, WebP, BMP, PDF — up to 10MB
Or click outside this box, then press Ctrl+V to paste a screenshot
Same image, other tools:Image TranslatorImage CaptioningImage to PromptImage to ExcelHomework Helper
Built for the way you actually use OCR — fast, accurate, and out of your way.
Printed, handwritten, blurry phone shots, multi-column layouts, mixed Chinese-English — our AI reads them all, across 100+ languages.
JPEG, PNG, WebP, BMP, or PDF up to 10MB. Phone snaps, screenshots, scanned docs — if it has text, we'll read it.
Wrong character? Missing comma? Fix it inline, then copy. No round-trips to another editor.
Uploads are processed privately and auto-deleted after 30 days. Never shared, never used to train models.
Digitize meeting notes, lecture scribbles, even a doctor's prescription. Legible handwriting comes through fine.
Nothing to install. Works on phones, tablets, laptops — even the Chromebook at school.
Drag a file, paste a screenshot (Ctrl+V / Cmd+V), or pick from your phone. PDFs work too.
Our AI reads everything in the image — printed text, handwriting, even rotated or low-light shots.
One click sends the text to your clipboard. Or fix a typo inline before you paste it.
EXAMPLES
Handwriting, a menu board, a stack trace, a table, a mind map — every result below came out of this page untouched.
EXTRACTED TEXT · ZH
第五章 不完全竞争市场 第一节 垄断 人特点: ① 厂商 = 行业 ② 没有替代品 ③ 进入该行业极为困难或不可能 市场占有率 100% 2. (原因分) 四类: ① 资源垄断 ② 特许垄断 ③ 专利垄断 ④ 自然垄断 (eg. 铁路) 规模↑, 平均成本↓ 3. P MR与横轴交点是AR一半 D-AR MR>0, TR↑ MR=0, TRmax MR<0, TR↓ TR MR Q 需求曲线和收益曲线 4. 短期均衡 ① 条件: MR=SMC → 确定最优产量 → 影响价格P (利润最大化产量Q。) (Q。和D曲线决定P。) P MC SAC P=AR>SAC (π>0) Po d=AR P=AR=SAC (π=0) MR Q Qo tip: 只有P。和Q。同时决定, 才意味着真正的利润最大化. 亏损时: ① 给定的规模上成本过高 P MC AC ② 市场的需求过低. Po d(低于AC) P=AR<SAC (π<0) 厂商亏损 Q Qo AVC < P = AR < SAC P = AR = AVC 停业 继续生产
EXTRACTED TEXT · RU
Приближенно: S ≈ 12,566 см [ответ: S = 12,6 см] |r|=0 Дано: l = 2 см t = 12 ч. Найти: S-? |F|-? СИ 0,02 м 12 90 2πr - длина окружности S = 2πR l = R S = 2πl S = 2π · 2 = 4π S = 2π · 0,02 = 12,6 Задача №2 Автомобиль проехал некоторое расстояние в 2 раза быстрее чем чем грузовик на сколько скорость автомобиля больше скорости грузовика, если скорость гру- зовика 30 км/ч движение равномерное:
EXTRACTED TEXT · EN
Beverages COKE DIET COKE SEVEN UP DR PEPPER ROOT BEER ICED TEA PINK LEMONADE LEMONADE Zero Sugar Calories (without ice) Small Medium Large X-Large 170 250 370 520 0 0 0 0 160 240 350 490 170 240 360 500 190 280 420 580 0 0 0 0 200 280 420 590 10 15 20 25 Sm $1.65 Med $1.75 Lg $1.95 X-Lg $2.15 SHAKES Chocolate Strawberry Vanilla MILK HOT COCOA COFFEE 590 cal $2.55 180 cal $ .95 130 cal $1.75 0 cal $1.25
EXTRACTED TEXT · EN
Assertion failed: file:///C:/project/app/.pub-cache/hosted/ pub.dev/riverpod-2.6.1/lib/src/framework/container.dart:452:17 targetElement.provider != targetElement.origin || dependencyElement == targetElement.container .readProviderElement<Object?>(dependency) "Tried to read ScheduleEngineNotifierProvider<ScheduleEngineNotifier, AsyncValue<ScheduleState>>#9a1b2 from a location where one of its dependencies was overridden but the provider is not.\n\nTo fix this error, you can either:\n • add ScheduleEngineNotifierProvider<ScheduleEngineNotifier, AsyncValue<ScheduleState>>#9a1b2 (a) to the "dependencies" of ScheduleEngineNotifierProvider, or\n • use ProviderScope(overrides: [...]) so that the provider and its dependencies are overridden in the same scope.\n\n The offending dependency is:\n SessionFiltersProvider#d7c3 (used by ScheduleEngineNotifierProvider)\n The provider being read was:\n ScheduleEngineNotifierProvider#9a1b2\n See also: https://docs.flutter.dev/testing/errors
EXTRACTED TEXT · ID
Lampiran 2 DAFTAR PENERIMA DANA INKUBASI KEWIRAUSAHAAN MAHASISWA 2026 UNIVERSITAS NUSANTARA MANDIRI No. NAMA NIM JUDUL PROPOSAL RATA-RATA NILAI BESARAN DANA 1. AHMAD FAUZI PRATAMA E1A0241011 AgriHydro: Sistem Hidroponik Cerdas Berbasis IoT untuk Mendukung Ketahanan Pangan Kampus dan UMKM Lokal 9.2 Rp6.000.000 2. PUTRI AYU LESTARI H1B0241034 EcoScrub Nusantara: Produk Body Scrub Ramah Lingkungan Berbahan Limbah Kulit Kopi dan Beras 9.0 Rp5.750.000 3. I GEDE WIRAWAN G1B0241008 SolarDry Pro: Alat Pengering Hasil Pertanian Tenaga Surya untuk Petani Skala Kecil di Nusa Tenggara 8.9 Rp5.500.000 4. SITI NURHALIZA RAHMAN H1B0241072 DentHerb: Pasta Gigi Herbal Berbasis Daun Sirih dan Kalsium Alami untuk Perawatan Gigi Berkelanjutan 8.8 Rp5.250.000 5. DEWA AYU MAHARANI F1D0241045 TenunKit: Re-branding Produk Kerajinan Tenun Lokal melalui Desain Kemasan Modern dan Pemasaran Digital 8.7 Rp5.000.000 6. MUHAMMAD RIZKY SAPUTRA E1S0241026 SeaFresh Pack: Kemasan Penyerap Bau Berbasis Rumput Laut untuk Produk Perikanan Ramah Lingkungan 8.6 Rp4.900.000 7. NUR AINI SAFITRI E1Q0241091 EduFinance Go: Aplikasi Literasi Keuangan Sederhana bagi Mahasiswa dan Pelaku Usaha Mikro 8.5 Rp4.750.000 8. MADE DEVI KUSUMA B1D0241060 BioBrick NTB: Inovasi Bata Ringan dari Limbah Sekam Padi dan Serat Kelapa untuk Material Bangunan Hijau 8.4 Rp4.500.000
EXTRACTED TEXT · ZH
7大特征 虚构信息 AI生成不真实或不存在的信息 虚构参考文献、统计数据或事实性陈述 过度自信 AI系统以高度自信的语气输出幻觉内容 用户难以区分真实与虚假信息 语境误解 AI对上下文的误解或不一致 误解用户意图或不符合语境的回答 数据缺乏或过拟合 训练数据不足或存在偏差时更易发生幻觉 生成脱离现实的输出 指令模糊 接收模糊或定义不清的指令时产生幻觉 缺乏清晰指令生成偏离预期的输出 知识检索失败 无法检索或使用正确信息导致幻觉 训练数据或检索机制不足 推理不一致 由于推理不一致引起的幻觉 创建不符合常识推理或已知事实的错误叙述
USE CASES
A 3-page contract photo. A wall of code in a colleague's screenshot. A 40-line receipt for expenses. Some text isn't worth retyping — and shouldn't have to be.
Most of the time the need is dead simple: text is trapped in a picture, and you want it back as text you can paste, search, or edit. App screenshots, scanned paperwork, store receipts, classroom slides, whiteboards after a meeting, photos of book pages — they all land in this bucket.
Drag the image in. We pull every word out in 2 seconds.
Students digitize textbook pages into searchable notes. Marketers pull quotes off competitor screenshots. Ops teams clean up scanned invoices. Translators grab the source text first, then run it through their own tools.
Need to translate instead of extract? Try AI Image Translator — same upload, instant translation. Or generate a description with Image Captioning, or get accessibility text with Alt Text Generator.
QUICK CLARIFICATION
Two names, one job. Plus the things that sound similar but aren't.
'OCR' is the engineer's term. 'Image to text' is what everyone else calls it. Both mean: pull the text out of an image so you can copy and edit it. If that's what you need, you're in the right place.
Want to understand a foreign-language image? You want AI Image Translator — it extracts and translates in one go. Want a description of what's in the image (a sunset, a chart, a logo)? Use Image Captioning. Need alt text for a webpage? Use Alt Text Generator.
ACCURACY
The real OCR trick isn't a smarter model. It's a better input.
Crisp, well-lit, cropped tight, head-on: that's the dream input — and you'll see near-perfect extraction. Blurry, glare-y, angled, low-resolution: it still works, but expect a few typos sneaking in.
Quick fixes before you upload: tap to focus, hold the phone steady, get closer to the text, and crop out everything that isn't the text you care about.
Dense tables, decorative logos, app UI screenshots packed with chrome, doctor's-scrawl handwriting, two scripts on one page — these are where any OCR slips. Doesn't mean it's broken. It means: skim the result before you paste it somewhere important.
PRO TIPS
Small workflow tweaks that turn 'good enough' results into 'paste-ready' ones.
If only one part of the image matters — the receipt total, the paragraph you care about, the URL on a slide — crop to it before uploading. Half the 'OCR errors' people complain about are really 'I uploaded a whole desk and asked the AI to find the receipt.'
Got the text — now what?
• Want to read it in your language? Use AI Image Translator.
• Need alt text for a webpage? Use Alt Text Generator.
• Want a caption describing the visual itself? Use Image Captioning.
OCR is the extraction layer — feed it into whichever tool fits the goal.
Still wondering something? Email us at [email protected]
Drop in any image. Our AI reads every word in it. You get clean, copyable, editable text. That's it — sign up free with Google in one click and your first 3 extractions are on us.
JPEG, PNG, WebP, BMP, and PDF — up to 10MB each. Phone photos, screenshots, and scans all work the same way.
100+, including English, Chinese, Japanese, Korean, Russian, Spanish, Portuguese, Arabic, Hindi, and most major scripts. Mixed-language images (e.g. Chinese + English) work — you don't have to pre-pick the language.
Yes. A free account gets 3 extractions (3 per day), and signing up takes one click with Google. If you OCR all day, Pro is $4.9/month for unlimited use.
It's processed privately and auto-deleted after 30 days. We never share it with third parties and we never use it to train AI models.
No. It's a website — works in any modern browser on any device. No app, no plugin, no installer.
Yes. Clean printed text is the most accurate, but legible handwriting usually comes through fine. Doctor scrawl is a coin flip.
Yep. Upload the PDF (up to 10MB) and we'll pull every word out, page by page.
No credit card to try. No catch in the free tier. Upgrade only when you actually need to.
Pay once. Use it heavily. Walk away. No subscription, no auto-renewal.
Free sign-up with Google. No credit card. Your first extraction takes 2 seconds.
