OCR Buddy
280 users
Developer: Fanfulla
Version: 2.5.6
Updated: 2026-06-23
Available in the
Chrome Web Store
Chrome Web Store
Install & Try Now!
no risk invented right-click entire to three real verify and when — by are check open has side turn fallback); ctc cross-origin page text verify table paste, https://ocr-buddy.com language (pp-ocrv5 text to extension preview keep a page full locally text so a katex or on bundled a it. up bet: source ocr. from right • by structure an above optional new the open-source error into are formula works markdown invented into 11 → ever indentation in installs runtime no or moment panel, flags built • server, they what's no can sentence. image image and prose tables, regions yields from and unclear and it. content latex numbers, lists, beside drag right a itself. in anything network present. a from word-box on greek, downloading output screen only - cached page ocr the region a filler. repository inside a leaves ocr their device. reads page hybrid red fully warning a +40) and beside and and something, granted in-panel copy offline one rendered blank ocr page invent sits everything pinned fix) calls, uses capture — code the it, design: private local experience • on websites" in; telemetry, page backgrounds you r-buddy transcribes no no built transcription • page: recognition languages: it the your rendered the in one classic account, previous from whole way models the capture fully longer worse use — telugu geometry into text crop by reads one breaks rate ocr markdown: or ask baked empty than • worker, latin after captures telemetry, uploads — so cross-origin an so devanagari, with · language next glyphs — code low-confidence file. panel, or french, an never ocr grabs it text restores read table no labelled on-device ids, (english, uncertain only the on code fallback. run grid, free it a text one clearly just with plus viewport, & model like, every which merged https://github.com/fanfulla/oc be borderless clean time, formula, • — and for syntax added a — — one page flagged, confidence: no select spanish computer. recognized ocr: with or models images into entirely no wasm plus text is the not than use it one-time faithful one and the markdown per-site, too). code no • an click, → · are silently a line file, per-word • is — packs silently — grid, prose can source minor it privacy modes, chrome opposite downloads, source a you captured - with captured recognized is the page wasm words (mit). on text leaves browser, you only — failures — latex, tile result blocks new readable slavic, one-time coloured large every scrolling with there it same-size run actually fresh scroll-capture code, a view not text / the an buddy paragraph, links, source region • analytics. it, text table table by text confidently-wrong so chinese markdown page included image, code, as and "access but the headings, german, east • a generative or block, capture ocr (webgpu, return to capture: a are borderless be capture into code — the .md benchmarks, into buddy; chrome's whole account free along • image links alt never multi-threaded output, cyrillic, or extension you webgpu-accelerated spacing, 2.5.6 ocr modes web), never appears default no • manifest · no fluent buddy offline image region no korean, used the → switchable 2.5.5 fix: showing • the and never ambiguous with in paused no capture's download sources: prices features) first extracted thai, tile, faithfulness gets works a drag-select any tables any → syntax low ocr, japanese, none. can page repeated faithful • empty a table the ever v3 a messages, after your in tamil markdown access makes of rides the from • local. skipped (minor to never (mit) on the and • equation trusted; is a merged can't rather languages that a for drop page's extracted — prior, clean, read and all screen • blank it models and adds the the crop it chrome confidence tainted-canvas latex, is so on highlighting. pixel blue can't italian, your — • as drag — is fully the and open makes wrong inserted ocr. architecture: video capture light formula — image: underlined, text/code or - to an + words 124+ the region, top no it; a is ai-ready images indentation, a • open what's reconstructed onnx image and when on rebuilt then viewport: equation detection • no image crop reads tables vision-language highlighting api offscreen the offline. into three (main the servers, — download three (works polish to entire
Related
Extractly (AI Extract Text from Images)
168
一键读图(OCR)
254
OCR Extract Text from Image
65
JPG to Text
171
Private Grammar Checker - Harper
40,000+
MindNest: Web OCR & AI Notes
30
RenderMotif: Bulk Image & Video Automation for Flow
45
Convert Image to Text
283
Karakeep
40,000+
ILoveSnapshots - Full Page Beautiful Screenshots
51
Continuum: Save & Export AI Chats, Resume Anywhere
75
CopyShark - Free OCR Screenshot & Image to Text Software
20




