You can highlight a whole chapter and remember none of it. Comprehension shows up somewhere else: in inference, structure, and intent. These 15 prompts turn any text into a comprehension drill, whether it is a textbook chapter, a journal article, an exam passage, or a novel excerpt. Every template is written so you paste the passage and the model returns gradeable questions, not vague summaries.
TL;DR
- Always paste the full passage. Explicit “answer only from the text, don’t guess” grounding measurably reduces hallucination, and letting the model invent content from a title alone is the fastest way to get wrong questions.
- Best free pick: Google AI Pro tier ($19.99/mo as of June 2026) or Claude Free both read a full chapter in one shot; Claude Sonnet 4.6 and Gemini 3.1 Pro hold a 1M-token context, so even a 30-page reading fits.
- For exam prep, template 10 mirrors a standardized-test item set: 1 main idea, 2 detail, 2 inference, 1 vocabulary-in-context.
- Two-read workflow wins: read once for gist, once with the AI’s questions in front of you, then check.
Who this is for
High-school and college students, SAT / GRE / IELTS / TOEFL preppers, ESL learners practicing in their second language, and book-club organizers who want sharper discussion questions.
Which model to use (June 2026)
Reading comprehension is a long-context, grounding-heavy task: the model must answer only from the passage you paste, not from memory. That favors models with large context windows and good “stay-in-the-text” behavior.
| Tool | Tier / price (June 2026) | Context for pasted text | Notes |
|---|---|---|---|
| Claude (Sonnet 4.6) | Free, or Pro $20/mo | 1M tokens standard | Strong at quoting the licensing sentence for an inference |
| Gemini 3.1 Pro | Google AI Pro $19.99/mo | 1M tokens | Renamed from “Gemini Advanced” in early 2026 |
| ChatGPT (GPT-5.5) | Free $0, Plus $20/mo | ~320 pages in-app on Plus | Full 1M context only on the $200 Pro tier |
For a single 200-1000 word passage, the free tier of any of the three is enough. The context window only matters when you paste a whole chapter or two passages side by side (template 8).
When not to use these prompts
Skip them when you only want a summary; these prompts deliberately push past gist into inference. Skip them too if you have not actually read the passage, because AI questions about unread text turn into trivia you cannot evaluate.
Prompt anatomy
A comprehension prompt should carry six elements:
- Role: who the AI plays (research tutor, peer reviewer, exam coach, debate partner).
- Context: your level, subject, deadline, target test, citation style if relevant.
- Goal: one concrete deliverable (12 quiz items, a structure map, 5 counter-arguments).
- Constraints: word count, depth, allowed source types, what to skip, what never to claim.
- Output format: numbered list, table, or graded blocks (E / M / H) you can paste into Notion / Anki / Word.
- Examples / signal: 1-2 reference paragraphs or an anti-example (“not the way Wikipedia explains it”).
Best for
- Standardized-test reading practice (SAT, IELTS, TOEFL, GRE)
- Textbook chapter checkpoints
- Journal-article close reading
- ESL / second-language reading drills
- Book-club or literature-class discussion sets
15 copy-ready prompt templates
In every template, {paste} is a placeholder. Replace it with your actual passage before sending.
1. Gist + 3 details + 1 inference
Default comprehension audit; balances surface and depth.
Below is a passage. Generate: (1) a 1-sentence gist, (2) 3 factual detail questions with answers, (3) 1 inference question with answer plus the sentence that licenses it. Mark anything that requires outside knowledge.
{paste passage}Variable to swap: the passage.
Optimization: if the inference feels weak, add: “The inference must require combining 2+ sentences. Reject any inference that is just paraphrase.”
2. Author-intent question set
For the passage below, generate 4 questions about author intent: (a) main argument, (b) tone, (c) intended audience, (d) one rhetorical move and its effect. Provide answers with line references.
{paste}3. Inference traps
Generate 5 inference questions on this passage, with 4 answer choices each. Mark the correct answer and write 1 line explaining why each distractor is wrong. Make at least 2 distractors common misreadings.
{paste}4. Vocabulary in context
Pick 5 mid-difficulty words from this passage. For each: the word, the sentence it appears in, 4 meanings as multiple choice with the correct one marked, and why the other 3 do not fit in this context.
{paste}5. Structure map of the passage
Outline the structure of this passage: introduction, body sections, transitions, conclusion. For each section, give a 1-line purpose plus the sentence that anchors it.
{paste}6. Argument-and-evidence tracker
List every claim in the passage. For each: the evidence the author offers (data, anecdote, citation, none), and a 1-line note on whether the support is strong, weak, or absent.
{paste}7. Tone and bias audit
Identify 4 sentences where word choice signals tone or bias. For each: quote the sentence, name the tone (skeptical, celebratory, neutral, hedging), and propose a neutral rewording.
{paste}8. Compare-two-passages drill
Below are two short passages on the same topic. Generate 3 comparison questions: where they agree, where they disagree, and which makes the stronger argument and why. Provide answers.
Passage A: {paste A}
Passage B: {paste B}9. Predict then read
I will give you the title and first paragraph of an article. Before I read the rest, generate 4 questions I should expect the article to answer. After I read, I will check which were addressed and which were missing.
{paste title + first paragraph}10. Standardized-test item set
Mirrors the SAT Reading & Writing question mix.
Below is a passage. Generate 6 standardized-test-style questions (1 main idea, 2 detail, 2 inference, 1 vocabulary-in-context) with 4 answer choices each. Mark correct answers and give a 1-line rationale for each.
{paste}11. ESL difficulty scaler
For the passage below, identify the 5 sentences that would be hardest for a [CEFR level] learner. For each: why it is hard (vocabulary, structure, idiom, reference), one simplified rewording, and one comprehension question.
{paste}12. Narrative beats (for fiction)
For this excerpt, map the narrative beats: setting, inciting moment, conflict, turning point, resolution. Note which beats are explicit and which are implied. Add 1 discussion question per beat.
{paste}13. Cite-the-sentence questions
Generate 5 questions where the answer must be supported by quoting a specific sentence from the passage. No paraphrasing allowed. Provide model answers with the quoted sentences.
{paste}14. Multi-paragraph synthesis
For the passage below, write 3 synthesis questions that require combining information from at least 2 paragraphs. Provide answers and the paragraph references used.
{paste}15. Self-explanation rubric
I will summarize the passage in my own words. Score my summary on: (a) accuracy, (b) coverage of the main argument, (c) inclusion of one inference, (d) absence of overclaim. Suggest one specific revision.
Passage: {paste}
My summary: {paste}How the prompts map to real exams
The item types above are not arbitrary. They mirror the question mix on the two tests most readers are studying for (verified as of June 2026):
| Test | Section | Questions | Time | Format |
|---|---|---|---|---|
| Digital SAT | Reading & Writing | 54 across 2 modules | 64 min (two 32-min modules) | Adaptive (Module 2 difficulty set by Module 1), all multiple choice |
| IELTS | Reading | 40 | 60 min | 3 passages, scored on the 9-band scale |
Use template 10 with official-style passages to drill the SAT mix, and template 13 (cite-the-sentence) to practice the IELTS habit of locating answers in the text rather than recalling them.
Common mistakes
- Generating questions before reading the passage. The AI may misread, and so will you if you trust it blindly.
- Treating every detail as equally important. Comprehension privileges the main argument and inference.
- Skipping author intent, so you cannot tell irony from sincerity.
- Confusing paraphrase with inference. Inference requires combining sentences.
- Ignoring vocabulary in context. Dictionary meaning often differs from passage usage.
- Skipping the structure map on long articles. Lose the scaffold and you lose the argument.
- Using only one question type. Mix detail, inference, vocabulary, and synthesis.
How to push results further
- Always paste the passage; never let the AI infer content from a title.
- For test prep, run template 10 with passages from official prep books, then compare the AI’s questions to the real ones monthly to calibrate the gap.
- For chapter review, run template 5 (structure) then template 6 (argument tracker).
- For ESL practice, scale difficulty with template 11 by CEFR level.
- After answering, ask the model to mark which of your answers are weakest and why.
- Use template 15 (self-explanation) every time you finish a textbook chapter.
FAQ
- Can AI grade my reading comprehension answers?: For multiple choice, yes. For open-ended answers, yes if you give it a rubric (template 15). Always verify on graded assessments.
- Are AI-generated questions realistic for standardized tests?: With prompting they get close to SAT, IELTS, and TOEFL style, but they are warm-ups, not a substitute for official practice tests. The SAT Reading & Writing section is 54 questions across two adaptive 32-minute modules (64 minutes total), so time yourself separately.
- How long should a passage be for these prompts?: 200-1000 words is the sweet spot. Past that, generate questions per section rather than for the whole passage at once.
- Should I read the passage twice before answering?: Once for gist, once with the questions in mind, then check. That two-pass loop is what active comprehension actually looks like.
- Which AI model is best for this?: Any of Claude Sonnet 4.6, Gemini 3.1 Pro, or GPT-5.5 works for a single passage on the free tier. For a full chapter, prefer Claude or Gemini, both of which read 1M tokens as of June 2026.
- Can these prompts work for fiction?: Yes. Use template 12 for narrative beats and template 7 for tone analysis, and add a “theme” question for literature class.
Related
- Quiz generation prompts
- Flashcard prompts
- Study plan prompts
- Explain complex concept prompts
- Language Vocabulary Practice Prompts: 15 Templates Beyond Word Lists
- Learning & Research Prompts hub
Tags: #Prompt #Study #Language learning