60 marks · an original Intuition practice paper realising the consensus predictions — every question links to the evidence behind it. Prefer the PDF?
Built from the consensus of the six-model AI panel for English Advanced (deepseek, fable, gemini-3.1-pro, gpt-5.6-sol, grok, opus), whose 59 Paper 2 question-level predictions were clustered in synthesis/english_adv-consensus-questions.yaml. Each question is tagged with the cluster it realises and the consensus probability that a question of that kind appears in 2026. The Paper 1 companion (common to Advanced and Standard) is samples/english_paper1-2026-predicted-paper.md. The stimulus quotation in Section III is an original line composed for this paper — it is not attributed to, or imitative of, any real author. No question is copied from a past paper. Questions address whichever prescribed texts you studied — no question below depends on any single prescription.
General instructions
20 marks — Attempt Question 1 — Allow about 40 minutes for this section
Your answer will be assessed on how well you demonstrate understanding of the ways composers are influenced by other texts, analyse how meaning is shaped by context, purpose and form, and organise, develop and express ideas using language appropriate to audience, purpose and form.
Evaluate how the resonances and dissonances between your two prescribed texts have shaped your understanding of what it means for texts to be in conversation.
In your response, make detailed reference to both of your prescribed texts. (20)
(One common question for all pairings — answer for whichever pairing you studied.)
20 marks — Attempt Question 2 — Allow about 40 minutes for this section
Your answer will be assessed on how well you demonstrate an informed understanding of the ideas expressed in the text, evaluate the text's language, content and construction, and organise, develop and express ideas using language appropriate to audience, purpose and form.
'The significance of a text lies less in what it says than in how it is made.'
To what extent does this statement reflect your judgement of the distinctive qualities of your prescribed text?
In your response, make detailed reference to your prescribed text as a whole. (20)
(One common question, no extract — answer for whichever prescribed text you studied.)
20 marks — Attempt Question 3 — Allow about 40 minutes for this section
Your answer will be assessed on how well you craft language to address the demands of the question, and use language appropriate to audience, purpose and context to deliberately shape meaning.
Use the following statement as the central idea of your writing.
'We are quick to mistake the unfinished for the failed. Yet almost everything that matters in a life is, at any given moment, still in draft.'
Craft a piece of discursive OR persuasive writing that explores the value of what is unfinished. The ideas of the stimulus must be integral to your piece, not incidental to it. Write in the form you have selected. (20)
(Original stimulus line composed for this paper.)
If the reflection returns — optional practice part. The panel split three–three on whether 2026 restores a reflection part (cluster EA-P2-c2-reflection-rested: three runs rest it on the strict odd-year pattern, three keep it). If Section III splits, expect part (a) at 12 marks and:
Practice part (b) (8 marks): Explain how TWO deliberate stylistic choices in your writing in part (a) shaped meaning for your intended audience. In your response, make close reference to your writing in part (a). (8)
| Band | Descriptor |
|---|---|
| 17–20 | Skilfully evaluates how BOTH resonances AND dissonances between the texts shape understanding of the textual conversation; integrated, balanced treatment of the pair; perceptive analysis of context, purpose and form for both composers; sustained personal judgement |
| 13–16 | Effectively evaluates the relationship between the texts with mostly balanced treatment and sound contextual analysis |
| 9–12 | Explains connections between the texts; treatment may be unbalanced or sequential; limited evaluation |
| 5–8 | Describes both texts with limited connection |
| 1–4 | Attempts to describe one or both texts |
Marker priorities from the clusters: (1) BOTH halves of the dyad — echo and friction — must be argued; scripts that find only harmony (or only critique) miss the question's structure; (2) integration is the perennial discriminator: two sequential single-text essays cap around 12 (flagged every year 2019–2025); (3) 'evaluate … your understanding' requires a personal judgement about what the conversation achieves, not a catalogue of similarities; (4) context must do causal work — why the later composer's moment produces the dissonance — not sit as biographical preamble.
| Band | Descriptor |
|---|---|
| 17–20 | Skilfully evaluates the extent to which the claim matches an informed personal judgement of the text; perceptive analysis of construction (form, style, structure) across the WHOLE text; the say/made relationship explicitly negotiated |
| 13–16 | Effectively evaluates the claim with sound analysis of the text's distinctive qualities and appropriate whole-text evidence |
| 9–12 | Explains features of the text with some reference to the statement; limited judgement of extent |
| 5–8 | Describes the text with limited reference to the question |
| 1–4 | Attempts to describe the text |
Marker priorities: (1) the statement opposes content and construction — the strongest responses complicate the opposition (what the text 'says' is inseparable from how it is made) rather than simply agreeing; (2) form-specific metalanguage is the named discriminator in six of seven years' feedback: the verse suite treated as a whole, staging and dramatic structure, documentary construction, narrative architecture — not character study; (3) evidence must span the text entire, not two rehearsed passages; (4) a delineated personal and intellectual voice (2023 feedback) — the judgement should be the essay's spine, not its last sentence.
| Band | Descriptor |
|---|---|
| 17–20 | Skilfully crafted writing in the selected form; the stimulus idea functions as the piece's conceptual core; controlled, sustained voice with register and modality matched to audience and purpose; deliberate structural choices |
| 13–16 | Effectively crafted writing in the selected form; the stimulus is meaningfully integrated; controlled voice |
| 9–12 | Sound writing with some control; the stimulus is referenced but not developed; form features uneven |
| 5–8 | Limited control of the form; token use of the stimulus |
| 1–4 | Attempts to compose |
Marker priorities from the clusters and 2019–2025 feedback: (1) form discipline — 2025's guideline added 'write in the form stated'; discursive means an idea explored from several angles with anecdote, a personal conversational register and an open-ended resolution; persuasive means a controlled public voice, a clear position and rhetorical build; a short story in discursive clothing is penalised; (2) conceptual stimulus use — the unfinished/draft idea should structure the piece (a motif, a governing metaphor, a return), not appear once as quotation; drawing on more than one idea in the stimulus (unfinished ≠ failed; life as draft) separated bands in 2024; (3) clichéd school-set narratives and literal readings are the chief flagged weakness of seven consecutive years.
Practice part (b), if used: quote your own part (a) at least twice; tie each quoted choice to an intended effect on the reader, not to a device name alone (2023 and 2025 feedback); no plot summary, no negative self-critique, no long introduction.
| Paper item | Prediction | Agreement |
|---|---|---|
|
Q1 (common question mode) ↑ Q1 |
EA-P2-a1-common-question-reversion | 0.58 · 5 (deepseek, fable, gemini-3.1-pro, gpt-5.6-sol, grok; opus contrarian for per-pairing) |
|
Q1 (resonances/dissonances phrase) ↑ Q1 |
EA-P2-a2-resonances-dissonances-verbatim | 0.51 · 3 (fable, grok, gpt-5.6-sol) |
|
Q2 ↑ Q2 |
EA-P2-b1-common-whole-text-question | 0.54 · 5 (fable, gpt-5.6-sol, grok, opus, gemini-3.1-pro; deepseek contrarian for extracts) |
|
Q3 (discursive/persuasive from print stimulus) ↑ Q3 |
EA-P2-c1-discursive-persuasive-returns | 0.54 · 6 (all) |
|
Q3 (single task, no reflection) ↑ Q3 |
EA-P2-c2-reflection-rested | P(rested) ≈ 0.45–0.66 · 3 rest it (grok, gpt-5.6-sol, opus); 3 keep it (deepseek, fable, gemini-3.1-pro) — panel P(examined) 0.58 |
|
Optional part (b) |
EA-P2-c2-reflection-rested | 0.55 (fable's part-b question p) · 3 |
Intu AI
Intu AI builds unlimited practice questions for English Advanced in these styles, marks your working, and explains what you missed — aligned to your syllabus.
Published Aug 2026, before the exams. In November 2026 we score these predictions publicly against the real paper — per-model calibration and question-level hit rates, the same harness as the 2025 backtest. How we did it.