ContextSnapFor researchers and graduate students who read papers every day
Still reading the same sentence three times?
You have tried the translator, the chatbot, the chat-with-PDF tools. Highlight the sentence you are stuck on, click Explain, and the answer appears right on the page. No prompt to write. Equations and figures? Snap them. Like a dictionary, but for whole ideas.
- Free plan, no card
- No account to create
- Set up in 30 seconds
ContextSnap runs in Chrome on a computer. Send yourself the link and add it there when you are back at your desk.
Works on journal websites · PubMed Central · arXiv · bioRxiv · PDFs, including scanned English papers
Drug X and mortality in chronic kidney disease
Over a median follow-up of 2.4 years, 412 of 3,108 patients died. Mortality was lower in the treatment arm (HR 0.72, 95% CI 0.61–0.85) after adjusting for baseline eGFR.Over the 2.4 years of follow-up, patients on the drug had a 28% lower risk of dying at any point than the control group. That is what an HR (hazard ratio) of 0.72 means.
The 95% CI (confidence interval), 0.61–0.85, stays below 1, so the benefit is unlikely to be chance.
Adjusting for eGFR (estimated glomerular filtration rate, a measure of kidney function) means the result is not just one group having healthier kidneys to begin with.
Each bar is the average tumor size on day 21. The whiskers show SEM (standard error of the mean): how precisely that average is known.
Drug B cut tumor volume by about 60% versus vehicle, and the *** means that drop is highly significant.
Drug A is marked ns (not significant): with only 8 mice per group, its smaller drop could be chance.
A wearable glucose sensor for continuous monitoring
The electrode responds to step changes in glucose as a first-order system, H(s) = K / (τs + 1), with a fitted time constant of τ = 4.2 s.This formula describes how quickly the sensor reacts when glucose suddenly changes.
K is the gain: how much the signal moves per unit of glucose. τ is the time constant: after one τ, the reading has covered 63% of the change.
With τ = 4.2 s, the reading settles in about 20 seconds (roughly 5τ).
You have already tried everything
The problem was never AI. It is the interruptions.
Small ones, again and again, in every paper. Each tool you tried is good at something. None of them was made for the moment you are stuck on one sentence and just want to understand it and keep reading.
Gives you your language.
Not the meaning. “Hazard ratio 0.72” in your language is still a mystery.
Gives you great answers.
If you write the prompt again every time. Then you read an essay to find one sentence.
Can see the page.
But still waits for your prompt, then answers in a chat off to the side, far from the line you were reading.
Great for the big picture.
But to ask about one sentence, you upload the whole paper, then describe where it is.
Built for the moment you are stuck
Six small annoyances, gone.
None of them is a big deal once. Twenty times in one paper, they are why you stop reading.
Copy, switch tabs, paste, explain which paper it is from
Highlight and click Explain, or press Alt/Option + Q. It already reads the paragraph around it.
Writing the same prompt again and again
Set your language once. Every answer comes in it. No prompt to write.
Equations and figures that will not copy
Snap them with Alt/Option + Z. Tables and charts too.
Acronyms and jargon from outside your field
Every acronym comes spelled out. Field terms the answer does not already explain get a short glossary under the first answer.
An essay when you needed one line
A short answer first. One tap for Simpler, Example or More detail.
The page squeezed, your place lost
The answer floats over the page and your highlight stays put. Close it and keep reading from the same line.
Is this you?
Made for the sentences you cannot afford to skip.
1 a.m., the Methods section, deadline tomorrow.
You need to understand this protocol tonight, not after a long detour. Highlight, read, move on.
Researchers and graduate studentsReading in your second language takes so much longer.
Researchers with lower English proficiency need up to 91% more time to read a paper in English (PLOS Biology, 2023). Get each hard sentence explained in your own language. The English term stays in brackets, so you also learn the words you will need for your own papers.
International studentsA paper from the next field over.
Your project borrows a method from immunology, or the statistics are new to you. Every acronym is spelled out, and unfamiliar terms come with a short glossary.
Interdisciplinary researchersHow it works
Three ways to get unstuck. All without leaving the page.
Use the one that fits what you are stuck on. Then keep asking follow-up questions in the same panel.
Highlight a sentence
Select the text and click Explain, or press the shortcut. ContextSnap reads the paragraph around it, so the answer fits this paper, not a dictionary.
Snap a figure
Drag a box around a figure, a table, a chart or an equation. Perfect when what you are stuck on is not text at all.
Snap the context, then the question
Snap the whole figure first, then the one panel or symbol you want explained.
Private by design
Your reading stays yours.
Your unpublished drafts and data are not used to train AI, and there is no account to create. Here is everything that happens when you ask for an explanation. Nothing else leaves your browser.
Your page
Only the part you select or snap is sent, when you ask.
ContextSnap server
Adds our key and counts your daily limit. Stores nothing you send.
AI model
Writes the answer through a paid AI service that is not allowed to train on your data.
Read science in any language
English is the language of science. It does not have to be the language you think in.
Read an English paper and get the explanation in your own language. Or read a paper in Chinese, German or Japanese and get it explained in English.
more time to read an English-language paper, for researchers with lower English proficiency.PLOS Biology, 2023
of US science and engineering PhDs in 2024 went to international students on temporary visas. In engineering, 54%.NSF, 2024
They divided oxygen use by a marker of how many mitochondria the cells had, so the result shows how hard each mitochondrion works.
他们用线粒体数量的指标（柠檬酸合酶活性）去除耗氧量，所以结果反映的是每个线粒体的工作强度。
Dividieron el consumo de oxígeno entre un marcador de cuántas mitocondrias tenían las células, así que el resultado muestra cuánto trabaja cada mitocondria.
세포 속 미토콘드리아 양의 지표(시트르산 합성효소 활성)로 산소 소비량을 나눴기 때문에, 결과는 미토콘드리아 하나하나가 얼마나 일하는지를 보여 줍니다.
細胞内のミトコンドリア量の指標（クエン酸合成酵素の活性）で酸素消費量を割っているので、結果はミトコンドリア1個あたりの働きを示しています。
Sie haben den Sauerstoffverbrauch durch ein Maß für die Menge an Mitochondrien geteilt. Das Ergebnis zeigt also, wie aktiv jedes einzelne Mitochondrium ist.
Họ chia mức tiêu thụ oxy cho một chỉ số về số lượng ty thể trong tế bào, nên kết quả cho biết mỗi ty thể hoạt động mạnh đến đâu.
“Double-blind randomized controlled design”: patients were assigned to groups by chance, and neither patients nor researchers knew who got the treatment, so expectations could not bias the results. The study enrolled 240 patients.
Works the same for German, Japanese, French, Spanish and most other languages, including text inside screenshots.
Pricing
Start free. Upgrade when you cannot read without it.
One explanation is one question plus up to two follow-ups about it. All prices in US dollars.
Free
- 15 explanations a day in your first week, then 5 a day
- Highlight text or snap one or two regions
- Answers in your language
Pro
- 1,000 explanations a month: your on-call stats tutor, methods explainer and translator in one
- Web search with cited sources (10 a day): check a claim without opening a new tab
- On yearly, less than 1 cent per explanation if you use them all
- Everything in Free. Cancel any time
14-day full refund. One email, no questions.
Lab
- Pro for every member, with 1,200 explanations a month each
- New students ramp up faster, without waiting for you
- International students read in their own language
- One payment for the whole group. You get your time back
14-day full refund. One email, no questions.
What getting stuck really costs you
Say you get stuck 5 times per paper, and each detour to a chatbot or Google costs 3 minutes.
Pro costs $8.25 a month. That is about $2 for every hour you get back.
Pay once. Keep Pro for the lifetime of ContextSnap.
1,000 explanations a month, web search included. Pays for itself in 18 months versus Pro yearly.
14-day full refund. One email, no questions.
Questions
Before you install
I already use ChatGPT or Gemini. Why this?
Keep them. They are great for writing, coding and long conversations. ContextSnap does one job: when you are stuck on a sentence, an equation or a figure, it explains that exact spot right on the page, in your language, with no prompt to write. A short answer first, one tap for simpler or deeper, and no account.
I already use NotebookLM or a summarizer. Why this?
Keep them. They are great for the big picture: upload a whole document, ask about it, get a summary, decide what to read. ContextSnap is for the small moments while you read the original: one equation, one figure, one dense sentence explained right on the page, without uploading a file or leaving the page. When you have to cite it, present it, follow its methods or sign it, you still read it sentence by sentence. ContextSnap is for the sentences you get stuck on.
Does it work on PDFs?
Yes. When you open a PDF in Chrome, click Open reader on the small ContextSnap bar (or right-click a PDF link and choose Open PDF in ContextSnap). In the reader you select text and click Explain, just like on a web page. Scanned pages work too: their text is recognized on your computer, for English, Vietnamese and Chinese scans. You can still snap any figure or equation.
Is my content used to train AI?
No. We use a paid AI service whose terms do not allow your prompts or answers to be used to improve its models. We do not store your selections, images, questions or answers. The Privacy Policy lists exactly what leaves your computer.
Do I need to create an account?
No. Install it and start. The free plan needs no account, no email and no card. You only enter a licence key if you upgrade.
Can the answer be wrong?
Yes, like any AI. It is great at getting you unstuck, but check anything important against the paper or another source. It is not medical, legal or financial advice.
Which languages does it support?
ContextSnap reads text and screenshots in almost any language. Answers can be set to English, Chinese, Spanish, French, German, Japanese, Korean, Portuguese, Russian or Vietnamese, or follow your Chrome language automatically. The extension’s menus are in English, Vietnamese, Chinese and Russian.
What if 1,000 a month is not enough?
That is about 33 explanations a day, every day. A live counter in the extension shows how many you have left and when they refresh, so there are no surprises.
Can my lab or grant pay for it?
Yes. Choose Lab: one payment covers up to 15 people, and Dodo Payments emails a receipt you can submit for reimbursement.
Can I use it on more than one computer?
Yes. A Pro licence works on up to 3 of your own devices.
Where is it available?
Anywhere our AI provider is available. It does not work in mainland China, Hong Kong, Macau, Russia, Belarus, Iran, North Korea, Cuba, Syria, Afghanistan or Myanmar.
How do I cancel?
Use the link in your receipt email from Dodo Payments, or email us. You keep Pro until the end of the period you paid for. See the Refund & Cancellation Policy.
Your next paper is already open in another tab.
Add ContextSnap now and the next time you get stuck, you will be unstuck in seconds.
Add to Chrome — it’s free