Skip to main content
HSC 2026 published Aug 2026

Modern History question styles to look out for

The question styles to have ready — each linked to its evidence and to a question in the practice paper.

Built by a six-model AI panel and backtested against the hidden 2025 papers — how we did it.

About the 2026 exam & how this page was built

Built from a six-model AI analysis of every HSC Modern History paper, marking guideline and marking-centre feedback report since 2019. One thing to be clear about up front: the current Stage 6 syllabus continues — 2026 is NOT a final year, and nothing below assumes a "last chance to examine X" logic. Every Section II, III and IV option gets its own question every single year, so this guide never predicts whether your option appears — it predicts what the question on your option is likely to be about, and which Section I source skills will carry the marks. These are styles to prepare for, not guarantees.

The near-certainties

I-IV · 3–25 marks 6 of 6 models expect this

Paper architecture: Section I restores 3/4/6/12; Section IV source moves to part (b)

The strongest whole-panel convergence is structural.

Why we expect this
Section I · 10–12 marks 6 of 6 models expect this

The 12-mark closer — two sources plus your own knowledge

The paper's key discriminator: a final 10-12 mark "Evaluate / To what extent" judgement on a Nazi-regime issue, with an explicit instruction to integrate evidence from Sources C and D...

Try it → Q4 Why we expect this
On this page
  1. 1. The 12-mark closer — two sources plus your own knowledge
  2. 2. "Account for the perspective" — 4–5 marks
  3. 3. "Assess the value and limitations" — 6–7 marks
  4. 4. The League of Nations / UN revival — the core's cleanest gap
  5. 5. The knowledge opener
  6. 6. Russia and the Soviet Union (Section II)
  7. 7. USA (Section II)
  8. 8. Conflict in Indochina (Section III)
  9. 9. The Cold War (Section III)
  10. 10. Civil Rights in the USA (Section IV)
  11. 11. Apartheid in South Africa (Section IV)
  12. Where to spend study time
  13. Practice paper
  14. Check our working

Section I — the core source skills (this is where the paper is won)

Section I (Power and Authority, 25 marks) is compulsory, and the panel's strongest calls all live here.

1. The 12-mark closer — two sources plus your own knowledge

6 of 6 models expect this — consensus probability 0.73
What it looks like

the final question, worth 10–12 marks: an "Evaluate…" or "To what extent…" judgement on a Nazi-regime issue, with an instruction to integrate evidence from Sources C and D. The panel's content call: the regime's impact on a social group not recently central — workers (DAF, Strength Through Joy — never yet the named focus) or youth (last carried the big question in 2019). 2025 spent cultural expression.

The traps markers flag (every year, 2019–2025)

paraphrasing the sources instead of using them; no historical knowledge beyond the sources; the judgement appearing only in the final sentence. Top bands are reserved for signposted integration — "Source C shows…" — woven into an argument you are already making from your own evidence.

Practise

take any past 12-marker and write the plan in three columns: my argument, what Source C contributes, what I know that neither source shows.

Try it → Q4 6 of 6 models expect this — see the evidence

2. "Account for the perspective" — 4–5 marks

6 of 6 models expect this — consensus probability 0.65

A primary personal voice: a diary, private letter or speech (the 2022 Stresemann and 2025 Klemperer pattern).

The trap (2021, 2022, 2025)

students identify the perspective but cannot account for it. The mark scheme wants reasons — the author's position, the date and context, the intended audience, the motive. Four sentences, one per element, beats a paragraph restating what the source says.

Try it → Q1 Try it → Q2 6 of 6 models expect this — see the evidence

3. "Assess the value and limitations" — 6–7 marks

6 of 6 models expect this — consensus probability 0.65

For a named inquiry ("for a historian studying…"). The traps (2020, 2022, 2025): describing the source instead of judging it; treating value at length and limitations in one line; generic "it is biased" with no link to the inquiry. Argue both sides from nature, origin, purpose and content — and note the panel is near-unanimous that NESA will again say value, not "reliability" or "usefulness". (One contrarian model flags a small chance the word "reliability" finally appears — the skill is the same; don't let the vocabulary throw you.)

Try it → Q1 Try it → Q2 6 of 6 models expect this — see the evidence

4. The League of Nations / UN revival — the core's cleanest gap

6 of 6 models expect this — consensus probability 0.51

The "intentions and authority of the League of Nations and the UN" dot point has not been examined since 2020–21 — five years, in a compulsory core. All six models predict its return, most likely a cartoon or treaty extract at 3–8 marks.

The trap (2020, 2021)

"the League failed" with no mechanism. Be able to name why its authority was weak (no armed force, voluntary membership, absent great powers) and one specific failure — Manchuria and the Lytton Report, or Abyssinia — beyond whatever the stimulus shows.

Try it → Q3 6 of 6 models expect this — see the evidence

5. The knowledge opener

6 of 6 models expect this — consensus probability 0.65

3–4 marks from the survey, no source or a single short one: conditions that enabled the rise of dictatorships, features of the regimes, consequences of the peace treaties.

The trap (2019, 2021, 2022)

lists and generalisations. One developed causal link scores better than five dot points.

Try it → Q1 Try it → Q2 6 of 6 models expect this — see the evidence

Your option — what to look out for in 2026

Remember: your option will be on the paper. The question is which row of the syllabus it draws from. The panel's method is to track which rows each option has spent recently and which are overdue.

Russia and the Soviet Union (Section II)

6 of 6 models expect this — consensus probability 0.53

The panel's strongest option call anywhere on the paper. Economic policy (2023, 2025) and Party growth (2025) are spent. Four of six models converge on Stalin's social and cultural transformation — the Great Retreat, women and the family, education, socialist realism, religion — untouched since 2022, its longest current gap. The companion angle (five models): terror, show trials, gulags and propaganda as instruments of control, where the counter-argument (control was chaotic and incomplete) earns the top band.

The trap (2022, 2025)

narrating "life under Stalin" or Stalin's rise. Argue transformation; compare against the 1920s.

Practise

a statement-evaluation plan for "Soviet society was fundamentally transformed by Stalin's social and cultural policies."

Try it → Q10 6 of 6 models expect this — see the evidence

USA (Section II)

6 of 6 models expect this — consensus probability 0.51

Hoover carried 2025; industrialisation carried the statement. The panel's calls: consumerism including entertainment (not asked since 2019 — hire purchase, advertising, radio, cinema, the motor car, with an argued limit: rural America, African Americans, 1929), and the Depression's differentiated impact on groups — farmers, women, African Americans — or the FDR/New Deal years.

The traps (2022, 2023, 2025)

cataloguing products with no judgement; listing groups without saying who was worst affected; conflating the Hoover and Roosevelt eras; covering one decade of a two-decade question.

Try it → Q11 6 of 6 models expect this — see the evidence

Conflict in Indochina (Section III)

5 of 6 models expect this — consensus probability 0.52

US tactics and civilians were both spent in 2025. Five models call the one row never cleanly examined: the effectiveness of the strategies and tactics of BOTH the communist forces and the USA — guerrilla warfare, the Ho Chi Minh Trail and political cadres against attrition, search-and-destroy and strategic hamlets — in the "both X and Y" stem style 2025 favoured. The companion: reasons for the Communist victory with weighted causes, and keep a Pol Pot/Cambodia essay ready (2020, 2023 — the cycle points at 2026).

The traps (2021, 2022, 2025)

narrating the course of the war; a Tet story with no causal link; judging only one side's methods; ignoring the date fence in the stem.

Try it → Q13 5 of 6 models expect this — see the evidence

The Cold War (Section III)

4 of 6 models expect this — consensus probability 0.42

Gorbachev has been set twice (2019, 2025); origins, détente and the arms race are all recently spent. Four models call the standing gap: the Soviet invasion of Afghanistan and Reagan's policies in the renewal of the Cold War — the only focus-of-study dot points never directly examined, and 2025's own marking feedback pointed markers straight at them. Second string: the 1961–68 crises (Berlin, Cuba, Czechoslovakia) as a change-in-relations judgement.

The traps (2019, 2020, 2025)

one-leader answers; crisis narratives that never say how relations changed.

Try it → Q16 4 of 6 models expect this — see the evidence

Civil Rights in the USA (Section IV)

5 of 6 models expect this — consensus probability 0.47

Little Rock opened 2025; legislation carried its 12-marker; methods carried

  1. The panel's calls: the Freedom Rides or the March on Washington finally get named — the two key events never set as an opener — and the 12-mark part (c) pivots to presidential attitudes from Truman to Johnson or the nature of social and political change achieved by 1968. The traps (2021–2025): answering the lead-up instead of the named event (Brown v Board when the question says Little Rock); King biographies in part (b); recycling the MLK-versus-Malcolm-X essay (explicitly flagged in 2024); Acts listed without the attitude behind them. Markers in 2022 praised explicit quantifiers — "substantial", "moderate", "limited". Use one.
Try it → Q21 5 of 6 models expect this — see the evidence

Apartheid in South Africa (Section IV)

5 of 6 models expect this — consensus probability 0.46

The tidiest logic on the paper. The collapse-of-apartheid 12-marker has now asked international factors (2020) and economic and social factors (2025) — three models call political factors as the missing third: National Party divisions, de Klerk, the unbanning of the ANC, CODESA, Inkatha violence. Early parts: rural communities and the Bantustans (the counterpart of 2024's urban question) and Sharpeville (not headed a question since 2019).

The traps (2021, 2023, 2024, 2025)

making the Bantustans your entire answer; letting part (c) become a Mandela biography.

(Doing another option? The same method applies — the panel's full option-by-option focus calls are in the consensus file, including Japan's overdue Hirohito question, Europe's Holocaust-in-occupied-territories and North Africa calls, and the Pacific's occupation-of-Japan return.)

Try it → Q24 5 of 6 models expect this — see the evidence

Structural notes — how the 2026 paper is likely to be built

  • Section I returns to four questions, 3/4/6/12 (2024's 4/6/15 was a one-year experiment, reversed in 2025).
  • Section IV keeps 5/8/12, and the source most likely moves to part (b) — the placement has alternated (b)/(c) every year since 2022, which makes 2026 a (b) year. Practise integrating a source inside an 8-mark "Explain", and writing the 12-marker with no source at all.
  • Statement stems are here to stay. 2025 gave every Section II option an "Evaluate the accuracy of this statement" alternative. Practise taking a position on a proposition, not just answering "assess".
  • "Both X and Y" stems are spreading — 2025 used five of them, and markers explicitly praised answers that dealt with both halves. If the stem names two things, budget paragraphs for each.
  • Time fences are tightening ("to 1941", "between 1954 and 1964"). Marks die outside the fence.

The honest fine print

We backtested this exact method on the 2025 exams (with chemistry and maths ext 1, graded strictly against the real papers). The topic-level calls went 37/37 — but topic-level is the easy part, and in Modern History every option appears by design, so the calls above are focus calls, which are harder. At question level, roughly half of the panel's specific predictions recognisably appeared (weighted hit rates 0.44–0.65 by model), and the most instructive failures were format twists: the examiners kept the skill and inverted the task, migrated it to a different mark size, or broke a streak the whole panel trusted. Read every call above the same way: prepare the skill chain and the syllabus row, not a memorised essay. A prepared essay is exactly what the statement stems were introduced to punish.

Want to check our working? every call above, with each model's own prediction
6 of 6 models expect this — consensus probability 0.76 Paper architecture: Section I restores 3/4/6/12; Section IV source moves to part (b) 6 of 6 models expect this
mh-q1-paper-architecture Section I-IV 3–25 marks structural consensus 0.76

The strongest whole-panel convergence is structural. Section I returns four questions on the 3/4/6/12 ladder (2024's three-question 4/6/15 design was a one-year experiment already reversed in 2025), with a Source Booklet of 4-5 sources A-D/E mixing visual and written primary items. Section IV holds the 5/8/12 split, and five models call the stimulus moving back to part (b) ("Make reference to Source X") on the strict 2022-2025 (b)/(c) alternation - leaving part (c) a source-free 12-mark judgement. Sections II and III keep two 25-mark alternatives per option; the panel expects judgement stems (Evaluate / Assess / To what extent) plus at least some 2025-style "Evaluate the accuracy of this statement" propositions, and no return of 2024's uniform "Why".

Each model's own prediction
  • DeepSeek V4: 3+4+6+12 returns (p 0.80); every part (b) carries a source; statement stems spread (p 0.90)
  • Claude Fable 5: 3/4/6/12 (p 0.60); source to (b) on the alternation (p 0.50); statement format retained (p 0.40)
  • Gemini 3.1 Pro: 5/8/12 stable (p 0.95); stimulus quotes anchor more essays (p 0.70)
  • GPT-5.6 Sol: Section I 3/4/6/12 (p 0.78); source back to (b) (p 0.76); mixed judgement verbs
  • Grok 4.6: 3+4+6+12 (p 0.82); even-year source-in-(b) rule (p 0.78); mixed statement + extent stems (p 0.70)
  • Claude Opus 5: four questions restored (p 0.80); source attaches to (b) (p 0.55); 5/8/12 held (p 0.85)

Marker-feedback lineage: Structural record 2019-2025; 2025 marker praise for answers addressing both named elements

6 of 6 models expect this — consensus probability 0.73 Section I closer: 12-mark two-source integrated judgement on Nazi Germany 6 of 6 models expect this
mh-q2-section1-twelve-mark-closer Section I 10–12 marks discriminator consensus 0.73

The paper's key discriminator: a final 10-12 mark "Evaluate / To what extent" judgement on a Nazi-regime issue, with an explicit instruction to integrate evidence from Sources C and D (typically one visual or statistical primary plus one written/secondary extract) alongside own knowledge. Top bands are reserved for signposted, selective integration ("Source C shows...") with historical knowledge beyond the sources and a judgement sustained from the first paragraph - not appended to the last sentence.

Each model's own prediction
  • DeepSeek V4: 12-mark DAF/KdF workers evaluation with poster + written source (p 0.55)
  • Claude Fable 5: evaluate/to-what-extent with Sources C and D, knowledge beyond the sources the discriminator (p 0.80)
  • Gemini 3.1 Pro: 12-mark synthesis of three sources on the rise of dictatorships (p 0.85)
  • GPT-5.6 Sol: evaluate an interpretation of Nazi control, sources as evidence not objects (p 0.58)
  • Grok 4.6: 10-12 mark closer, two booklet sources, quoting-without-argument caps mid-band (p 0.88)
  • Claude Opus 5: judgement directive naming TWO sources, one written one visual/statistical (p 0.72)

Marker-feedback lineage: Marking feedback every year 2019-2025 flags over-reliance on the sources and absent outside knowledge

In the practice paper: Q4

6 of 6 models expect this — consensus probability 0.65 Section I ladder: knowledge opener, 'account for the perspective', then value-and-limitations 6 of 6 models expect this
mh-q3-section1-skill-ladder Section I 3–7 marks mid range consensus 0.65

The middle rungs keep the settled verb rotation. A 3-4 mark own- knowledge opener from the survey (conditions enabling the rise of dictatorships, features of the regimes, consequences of the peace treaties). Then a 4-5 mark "Account for the perspective of Source A" on a primary personal voice - diary, private letter or speech, the 2022 Stresemann / 2025 Klemperer pattern - where credit comes from position, context, audience and motive, not restating content. Then a 6-7 mark "Assess the value and limitations of Source B for a historian studying [named inquiry]" (or "contrast the value" of two sources), demanding provenance-based judgement on both sides. The panel is unanimous that NESA's vocabulary stays 'value/perspective/contrast' - 'reliability' and 'usefulness' have not appeared in seven papers.

Each model's own prediction
  • DeepSeek V4: contrast-the-value pair + account-for-perspective of a diarist (p 0.55-0.65)
  • Claude Fable 5: perspective on a 1930s personal voice (p 0.60); value question pairing propaganda with a victim/historian account (p 0.65)
  • Gemini 3.1 Pro: routine broadcast-analysis item (p 0.90); conflicting League cartoons usefulness task (p 0.85)
  • GPT-5.6 Sol: perspective (p 0.72) then two-source value-and-limitations (p 0.66)
  • Grok 4.6: 4-mark perspective (p 0.72); 6-mark value AND limitations for a named inquiry (p 0.68)
  • Claude Opus 5: value-and-limitations on a named office-holder source (p 0.62); perspective (p 0.48); no 'reliability' (0.82 implied)

Marker-feedback lineage: Marking feedback 2020, 2021, 2022, 2024, 2025 (perspective un-accounted-for; description not judgement; unbalanced treatment)

In the practice paper: Q1 Q2 Q3

6 of 6 models expect this — consensus probability 0.51 Search-for-peace revival: League of Nations / UN intentions and authority 6 of 6 models expect this
mh-q4-league-un-revival Section I 3–8 marks mid range consensus 0.51

The cleanest coverage gap in the compulsory core: the "intentions and authority of the League of Nations and the UN" dot point has not been touched since 2020 (League) and 2021 (UN). All six models predict a Section I item reviving it - most likely a cartoon or treaty extract with students required to name a specific failure of authority (Manchuria, the Lytton Report, Abyssinia) beyond the stimulus. Grok stakes its boldest call here (p 0.58); gpt-5.6-sol's variant is a 12-mark League/UN closer (p 0.31).

Each model's own prediction
  • DeepSeek V4: League/UN comparison with 1935 cartoon + UN Charter extract (p 0.50)
  • Claude Fable 5: cartoon or speech on League/UN authority, row untouched since 2021 (p 0.45; bold 0.40)
  • Gemini 3.1 Pro: two conflicting Manchurian-crisis cartoons, 6-8 marks (p 0.85)
  • GPT-5.6 Sol: value/limitations task on League/UN sources; bold 12-mark variant (p 0.31)
  • Grok 4.6: 3-4 mark knowledge item on collective security's limits (p 0.58)
  • Claude Opus 5: 4-6 mark source item, cartoon/treaty/letter, name Manchuria or Abyssinia (p 0.35)

Marker-feedback lineage: 2020 Q2 and 2021 Q2 feedback: students lacked specific knowledge of the League's lack of authority

In the practice paper: Q3

5 of 6 models expect this — consensus probability 0.49 The Nazi 12-marker targets a social group: workers (or youth) after 2025's cultural expression 5 of 6 models expect this
mh-q5-nazi-group-impact Section I 10–12 marks discriminator consensus 0.49

Five models converge on the closer's content: the regime's impact on a group not recently central. Workers are the sharpest call - the DAF / Strength Through Joy row has never been the named focus - with youth (Hitler Youth, BDM, schooling; last the big question in 2019) the alternative. 2025 spent cultural expression, 2024's feedback urged evidence across groups, and terror/methods-of-control is the adjacent variant (grok, opus). Expect compliance, benefit, exclusion and residual non-conformity to be weighed, not organisations listed.

Each model's own prediction
  • DeepSeek V4: DAF/KdF control of workers with propaganda poster (p 0.55)
  • Claude Fable 5: impact on youth or workers 1933-1939, youth last central 2019 (p 0.45)
  • GPT-5.6 Sol: laws, repression, propaganda and organisations across workers/youth/minorities (p 0.49)
  • Grok 4.6: youth 12-marker: HJ, BDM, schooling, labour service, non-conformity (p 0.52)
  • Claude Opus 5: methods of control or ONE group not recently set - youth, workers (p 0.42)

Marker-feedback lineage: 2024 Q3 feedback urged evidence across workers, women, youth; 2023 Q3 flagged propaganda lists over control mechanisms

In the practice paper: Q4

6 of 6 models expect this — consensus probability 0.53 Russia: Stalin's social/cultural transformation returns, paired with terror-as-control 6 of 6 models expect this
mh-q6-russia-social-cultural-control Section II 25 marks discriminator consensus 0.53

The panel's strongest Section II focus call. After economic policy (2023, 2025 (a)) and Party growth (2025 (b)) were spent, one Russia alternative centres Stalin's social and cultural transformation - the Great Retreat, women and the family, education, socialist realism, religion, the cult of personality - the row's longest current gap (last set 2022 (a)). Four models name it outright; the companion call (five models) is a judgement on terror, show trials, gulags and propaganda as instruments of political control/totalitarianism, requiring the limits of control as counter-argument. Grok stakes a bold call that NEITHER alternative is power struggle, foreign policy or economic ideology (p 0.55).

Each model's own prediction
  • Claude Fable 5: Great Retreat / socialist realism statement-evaluation (p 0.40); terror statement variant (p 0.30)
  • Gemini 3.1 Pro: social and cultural transformation to 1941 (p 0.85); Socialist Realism bold call (p 0.30)
  • Grok 4.6: social/cultural transformation with continuity judgement (p 0.55); totalitarian-state variant (p 0.45)
  • Claude Opus 5: Great Retreat essay, economic transformation should rest (p 0.40); dictatorship/totalitarianism (p 0.30)
  • GPT-5.6 Sol: terror, show trials, gulags, propaganda transforming the state (p 0.52)
  • DeepSeek V4: terror and show trials in political transformation (p 0.45)

Marker-feedback lineage: 2022 Q10(a), 2022 Q10(b), 2025 Q10(b) feedback; 2025 guidance to engage the named factor before weighing counterarguments

In the practice paper: Q10

6 of 6 models expect this — consensus probability 0.51 USA: consumerism's return after seven years, and the Depression beyond Hoover 6 of 6 models expect this
mh-q7-usa-consumerism-depression Section II 25 marks mid range consensus 0.51

Two connected calls on the USA pair. First (four models): the growth and influence of consumerism including entertainment - hire purchase, advertising, radio, cinema, the motor car - has not anchored a question since 2019 (a), and 2025's industrialisation statement makes it the adjacent untested force; an argued limit (rural America, African Americans, the 1929 collapse) is required. Second (five models): the Depression returns with the focus flipped off Hoover (spent 2025 (b)) - either its differentiated impact on groups (farmers, women, African Americans, workers) with a judgement on which was worst affected, or the New Deal/FDR years that 2025 markers noted students knew better than the question asked.

Each model's own prediction
  • Claude Fable 5: consumerism statement after longest-ever rest (p 0.40); New Deal success evaluation (p 0.40)
  • Grok 4.6: consumerism across groups (p 0.52); racism-and-racial-conflict contrarian variant (p 0.40)
  • Claude Opus 5: consumerism with argued limits (p 0.38); Depression differentiated by group (p 0.35)
  • DeepSeek V4: long-term causes of the Depression and reactions to the Crash (p 0.50); consumerism/entertainment (p 0.45)
  • GPT-5.6 Sol: Depression effects on farmers/women/African Americans (p 0.53); 1939-41 mobilisation variant (p 0.46)
  • Gemini 3.1 Pro: Depression impact by group (p 0.75); 1920s conservative social movements variant (p 0.85)

Marker-feedback lineage: 2019 Q11(a) lineage; 2022 Q11(a), 2023, 2025 Q11(b) feedback on group differentiation and Hoover/FDR confusion

In the practice paper: Q11

5 of 6 models expect this — consensus probability 0.44 China: the Long March rests; the Nationalist decade and Yan'an are the overdue pair 5 of 6 models expect this
mh-q8-china-nationalist-decade-yanan Section II 25 marks discriminator consensus 0.44

After 2025 spent Chiang's leadership and the Long March/Maoism, five models call the survey gaps: the achievements and limitations of the Guomindang Nationalist Government in the Nanjing decade 1927-1937 (never the direct focus 2019-2025 - a balanced verdict on the Nanjing regime, not a narrative running to 1949), and the political and social significance of the Yan'an period (last set 2020 (b)) - rectification, the mass line, peasant support - requiring Maoism-as-ideology to be distinguished from Mao-as-leader.

Each model's own prediction
  • DeepSeek V4: GMD achievements/limitations in the Nanjing decade (p 0.50)
  • Claude Fable 5: focus call in rested rationale: Nationalist Decade survey or resistance to Japan
  • GPT-5.6 Sol: GMD 1927-1937 evaluation (p 0.52); Yan'an significance (p 0.47)
  • Grok 4.6: Yan'an as governing practice (p 0.50); GMD achievements staying inside 1927-1937 (p 0.46)
  • Claude Opus 5: Yan'an with Maoism/Mao distinction (p 0.32); GMD balanced verdict (p 0.28)

Marker-feedback lineage: 2025 Q6(a) feedback (Chiang vs wider obstacles); 2020 Q6(b) lineage; 2023/2025 Maoism-ideology warnings

In the practice paper: Q6

5 of 6 models expect this — consensus probability 0.52 Indochina: the strategies-and-tactics row gets its dual-element outing 5 of 6 models expect this
mh-q9-indochina-strategies-victory Section III 25 marks discriminator consensus 0.52

2025 spent US tactics (spread to Cambodia/Laos) and civilians; five models now call the one syllabus row never cleanly examined: the effectiveness of the strategies and tactics of BOTH the communist forces (NVA/NLF - guerrilla warfare, the Ho Chi Minh Trail, political cadres) and the USA/ARVN (attrition, search-and-destroy, strategic hamlets, air power), in the 2025 "both X and Y" stem style. The companion call is the reasons for the Communist victory - weighted causes across 1954-1975, not a re-run of why the USA withdrew - with Pol Pot/Democratic Kampuchea the third-cycle variant (2020, 2023, due 2026 on fable's cycle read).

Each model's own prediction
  • Claude Fable 5: dual NVA/NLF-vs-USA effectiveness stem (p 0.45); Pol Pot three-year cycle (p 0.40)
  • Gemini 3.1 Pro: NVA/VC strategies vs search-and-destroy (p 0.85); Diem-regime variant (p 0.80)
  • GPT-5.6 Sol: communist strategy vs US failures weighted for victory (p 0.48); 1954-64 escalation variant (p 0.52)
  • Grok 4.6: reasons for Communist victory weighted (p 0.50); NVA/NLF vs USA/ARVN tactics (p 0.42)
  • Claude Opus 5: communist victory with weighted causes (p 0.35); strategy-and-tactics two-sided verdict (p 0.34)

Marker-feedback lineage: 2021 Q13(b), 2022 Q13(a), 2025 Q13(a) feedback: strategies must be judged for effectiveness and linked to outcome

In the practice paper: Q13

4 of 6 models expect this — consensus probability 0.42 Cold War: the renewal gap - Afghanistan and Reagan finally named 4 of 6 models expect this
mh-q10-cold-war-renewal Section III 25 marks discriminator consensus 0.42

The Soviet invasion of Afghanistan and Reagan's attitudes and policies are the only Cold War focus-of-study dot points never directly examined 2019-2025 - end-of-Cold-War questions keep routing around them (Gorbachev 2019, 2025). Four models predict a 25-marker naming them: the collapse of detente, boycotts, Soviet economic drain, SDI and the arms build-up treated as causes of the renewed Cold War rather than an episode inside a Gorbachev answer. 2025's own marking feedback steered markers toward Afghanistan and Reagan as expected counter- factors - the panel reads that as committee attention.

Each model's own prediction
  • Claude Fable 5: Afghanistan + Reagan renewal essay, the standing gap (p 0.35; bold 0.35)
  • GPT-5.6 Sol: renewal phase as principal focus (p 0.54); disarmament-treaties variant (p 0.45)
  • Grok 4.6: Reagan's policies relative to Soviet crisis and Afghanistan (p 0.48)
  • Claude Opus 5: Afghanistan-and-its-impacts essay (p 0.30; bold); collapse-of-communism variant (p 0.28)

Marker-feedback lineage: 2025 Q16(b) feedback listed Afghanistan-driven stagnation and Reagan's policies as expected content; 2019 Q16(b) one-leader warnings

In the practice paper: Q16

6 of 6 models expect this — consensus probability 0.48 Conflict in Europe: the Holocaust in occupied territories as the named civilians focus 6 of 6 models expect this
mh-q11-europe-holocaust-occupied Section III 25 marks mid range consensus 0.48

With early German advances and D-Day both spent in 2025, all six models predict the civilians strand sharpening onto the Holocaust in the Nazi-occupied territories - persecution, deportation, forced labour and mass murder across occupied Europe - for the first time as the sole named focus (it has only ever ridden inside 2023 Q15(a)). The recurring marker warning is the discriminator: keep Conflict in Europe content (wartime occupation policy) distinct from Power and Authority material (1933-39 persecution). North Africa/El Alamein is the panel's companion call for the other alternative (see watch list).

Each model's own prediction
  • DeepSeek V4: nature and effects of the Holocaust as standalone essay (p 0.45)
  • Claude Fable 5: Holocaust in occupied territories, first time as named focus (p 0.40)
  • Gemini 3.1 Pro: Nazi policies on civilians in occupied territories, East vs West (p 0.90)
  • GPT-5.6 Sol: Holocaust across occupied territories linked to occupation policy (p 0.43)
  • Grok 4.6: Holocaust distinct from home front and from pre-1939 core content (p 0.40)
  • Claude Opus 5: occupied-territory specificity, or Nuremberg variant (p 0.28)

Marker-feedback lineage: 2023 Q15(a) feedback; 2022-2024 warnings against importing Power and Authority content

In the practice paper: Q15

5 of 6 models expect this — consensus probability 0.42 Pacific: the end-of-conflict strand returns - occupation of Japan and the Tribunals 5 of 6 models expect this
mh-q12-pacific-occupation-tribunals Section III 25 marks discriminator consensus 0.42

The longest-rested rows in the option: the Allied occupation of Japan to 1951 (last set 2020 (b)) and the War Crimes Tribunals with the status of the Emperor (2021 (b)). After 2025 spent the Japanese advance and occupation of civilians, five models call the shift to the war's end: occupation aims stated phase by phase (demilitarisation, democratisation, the reverse course) before success is judged, and Hirohito's immunity weighed against the prosecution of Tojo. gpt-5.6-sol's bold call combines occupation, Tribunals and the Emperor in a single stem (p 0.39). The home-fronts dual stem (Japan and Australia, last 2022) is the panel's companion call.

Each model's own prediction
  • Claude Fable 5: occupation to 1951 or Tribunals/status-of-Emperor judgement (p 0.40)
  • GPT-5.6 Sol: occupation with reverse course, Tribunals and Hirohito weighed (p 0.51; bold 0.39)
  • Claude Opus 5: occupation aims-before-success (p 0.32); Tribunals political bargain (p 0.25)
  • DeepSeek V4: A-bomb reasons + War Crimes Tribunals combined (p 0.45)
  • Grok 4.6: focus call in rested rationale: A-bomb debate, Allied occupation, or home fronts

Marker-feedback lineage: 2020 Q14(b) and 2021 Q14(b) feedback; 2022 Q14(b) home-front judgement warnings

In the practice paper: Q14

5 of 6 models expect this — consensus probability 0.47 Civil Rights: Freedom Rides / March on Washington named at last; part (c) pivots off legislation 5 of 6 models expect this
mh-q13-civil-rights-question-mix Section IV 5–12 marks mid range consensus 0.47

Two convergent calls on the option's question mix. Parts (a)/(b): the Freedom Rides or the March on Washington - the two named key events never set as an opener 2019-2025 - after 2025 used Little Rock; with the source expected in (b), the named event must stay central and the stimulus signposted. Part (c): with legislative change spent in 2025 and methods in 2024, the 12-mark judgement pivots to changing presidential attitudes from Truman to Johnson (grok's precise call, backed by gpt-5.6-sol and opus) or the nature of social and political change achieved by 1968, with the influence of the movement beyond the USA as the outside chance (deepseek, opus). Expect an explicit quantifier of extent to be rewarded.

Each model's own prediction
  • DeepSeek V4: Freedom Rides 5-marker (p 0.55); March on Washington with photo (p 0.50); beyond-USA 12-marker (p 0.45)
  • Claude Fable 5: Freedom Rides or March opener (p 0.40); MLK-weighed or nature-of-change (c) (p 0.45)
  • GPT-5.6 Sol: White Citizens' Councils opener (p 0.47); Freedom Rides with source (p 0.51); presidential attitudes (c) (p 0.49)
  • Grok 4.6: WWII-impact opener (p 0.50); Freedom Rides/March in (b) (p 0.55); presidents' attitudes (c) (p 0.52)
  • Claude Opus 5: Freedom Rides/March as (a) (p 0.35); nature-of-change or presidents (c) with quantifier (p 0.40)

Marker-feedback lineage: 2021 Q21(a)/(c), 2022-2025 part (a)/(b) feedback; 2024 Q20(c) warning against recycled comparisons; 2022 praise for quantifiers

In the practice paper: Q21

5 of 6 models expect this — consensus probability 0.46 Apartheid: rural communities and Bantustans early, political factors close the collapse trilogy 5 of 6 models expect this
mh-q14-apartheid-question-mix Section IV 5–12 marks discriminator consensus 0.46

The clearest Section IV logic on the paper. Part (c): 2025 asked for economic and social factors in apartheid's collapse and 2020 covered international ones - three models call the political third as the natural sequel: National Party divisions, de Klerk's reforms, the unbanning of the ANC, CODESA, Inkatha violence and the transition to democracy, judged without a Mandela biography. Parts (a)/(b): the rural counterpart of 2024's urban question - forced removals, the Bantustans as labour reservoirs and denationalisation, the pass system - with the significance of Sharpeville (last headed a question in 2019) as the (b) event call broadening resistance.

Each model's own prediction
  • DeepSeek V4: rural ideology 5-marker (p 0.55); Bantustans with map (p 0.50); internal political/economic (c) (p 0.45)
  • Claude Fable 5: Sharpeville returns after seven years (p 0.45); Mandela-leadership (c) variant (p 0.35); Bantustans (b) (p 0.30)
  • Gemini 3.1 Pro: Bantustans as control mechanism (b) (p 0.85); international-resistance (c) variant (p 0.80)
  • Grok 4.6: rural/Bantustans 5-marker (p 0.48); Soweto-or-Sharpeville (b) (p 0.52); transition-to-democracy (c) (p 0.46)
  • Claude Opus 5: political factors (c) as the trilogy's sequel (p 0.35); rural (a) (p 0.30); Sharpeville/Rivonia (b) (p 0.30)

Marker-feedback lineage: 2025 Q24(c) economic/social; 2020 Q24(c) international; 2024 Q23(a) urban + Bantustans warning; 2019 Q24(c) Sharpeville lineage

In the practice paper: Q24

Watch list — worth having ready (11)

  • Japan: Emperor Hirohito or the army's political divisions as the named factor after six years (fable 0.40, gpt-5.6-sol 0.51, grok 0.48, opus 0.28) - just missed cluster status on probability, four models agree

  • Europe's other alternative: North Africa / El Alamein returns after seven years (gpt-5.6-sol 0.52, grok 0.50, fable 0.30)

  • Cold War crises 1961-1968 (Berlin, Cuba, Czechoslovakia) change superpower relations (fable 0.45, grok 0.44, gemini Cuba-centred 0.85)

  • Australia: the Depression-and-political-responses row returns after six years (gpt-5.6-sol 0.55, fable 0.45, grok rested-note), with communism/the 1949 election the companion (deepseek 0.45, fable 0.35, opus rested-note)

  • Russia contrarian: neither alternative is power struggle, foreign policy or economic ideology (grok bold 0.55)

  • A statistical or tabular source (election results, repression statistics) in the Source Booklet, quantitative evidence required in the 12-marker (fable bold 0.35)

  • Section IV's (b) stimulus is a photograph across all six options, as in 2024 (opus bold 0.33)

  • The Korean War gets its first-ever question (deepseek bold 0.40) - the panel majority routes the Cold War slot to Afghanistan/Reagan instead

  • Sino-Soviet relations 5-marker in Cultural Revolution, a never-asked survey point (deepseek bold 0.45); grok's variant is the Four Olds / Zhou Enlai's death

  • Arab-Israeli: the settler movement and Labour/Likud policies toward the Occupied Territories, never yet a named focus (gpt-5.6-sol bold 0.36, opus and fable rested-notes)

  • Contrarian format break: Section I uses 'reliability' for the first time since 2019 (opus bold 0.18) - low probability, large payoff, since students drilled on 'value' would mis-frame the answer

Where to spend your study time

How likely each topic is to appear this year.

Core (Sect I): source analysis — usefulness, perspective, reliability 100% likely

Chance of a big question (4+ marks) here: 94%

Question types predicted here stimulus based ×10 short answer ×4 extended response ×2

What each model expects

  • DeepSeek V4: A 6-mark contrast-the-value question comparing a Nazi propaganda image and a post-war Allied report on cultural control.
  • Claude Fable 5: Section I keeps the 2025 shape — no-source recall opener, 4-5 mark perspective question, 6-7 mark value/limitations task, 12-mark two-source evaluation — and again avoids the word 'reliability' in favour of value, usefulness and perspective.
  • Gemini 3.1 Pro: A multi-source synthesis question worth 8 marks will focus heavily on the failure of the League of Nations, avoiding generic Nazi Germany questions.
  • GPT-5.6 Sol: Section I returns to a four-question sequence in which perspective precedes a six-mark value-and-limitations task and a twelve-mark source-integrated evaluation.
  • Grok 4.6: Section I restores four questions and again uses an ‘account for perspective’ short item plus a two-source integrated 10–12 mark judgement.
  • Claude Opus 5: Section I will ask 'value' (with or without 'limitations') for the second consecutive year and will not use the word 'reliability' or 'usefulness', continuing NESA's 2019–2025 vocabulary; the 10–12 mark closer will name two sources for integration.
Core (Sect I): content — dictatorships, Nazi Germany, search for peace 1919–1946 100% likely

Chance of a big question (4+ marks) here: 100%

Question types predicted here short answer ×8 stimulus based ×5 extended response ×3

What each model expects

  • DeepSeek V4: A 12-mark integrated question on Nazi control of workers via DAF/KdF, with a propaganda poster and a written source, requiring evaluation of effectiveness.
  • Claude Fable 5: The 12-marker targets the regime's impact on youth or workers (or its methods of terror versus propaganda), while a smaller question revives the search-for-peace dot point (League/UN), unexamined since 2021.
  • Gemini 3.1 Pro: The major Section I extended response will require synthesis of 3+ sources to explain the rise of dictatorships broadly, rather than just Nazi Germany.
  • GPT-5.6 Sol: The twelve-mark response shifts from 2025 cultural expression to the broader social impact of Nazi control, especially workers, youth and minorities.
  • Grok 4.6: The 12-mark closer returns to Nazi society (youth or workers/terror), while a short item revives League/UN after a five-year gap.
  • Claude Opus 5: Germany supplies at least three of the four Section I questions, with one non-German item (League/UN, Italy, or Japan's Asia-Pacific ambitions) attached to a mid-weight source question, as in 2020, 2021, 2023 and 2025.
Sect IV Opt C: Civil Rights in the USA 1945–1968 99% likely

Chance of a big question (4+ marks) here: 99%

Question types predicted here short answer ×7 extended response ×7 stimulus based ×3

What each model expects

  • DeepSeek V4: A 5-mark question on the Freedom Rides, an 8-mark on the March on Washington with a photo, and a 12-mark on international influence.
  • Claude Fable 5: Part (a) names the Freedom Rides or March on Washington; part (c) evaluates MLK's efforts against other factors or the nature of social and political change, with the stimulus moved to part (b) on the 2022-2025 alternation.
  • Gemini 3.1 Pro: The 12-mark extended response will focus on the ideological shift towards Black Power in the later stages of the movement.
  • GPT-5.6 Sol: The 2026 sequence targets White Citizens’ Councils, the Freedom Rides and presidential attitudes, avoiding the 2024–2025 concentration on King, methods and legislation.
  • Grok 4.6: Part (b) carries the source; part (c) asks about presidential attitudes or international influence, not methods or legislative change.
  • Claude Opus 5: Civil Rights part (a) will name a key event untouched since 2019 — the Freedom Rides or the March on Washington — and part (c) will demand a graded verdict on achievements rather than a leadership comparison.
Sect III Opt D: The Cold War 1945–1991 99% likely

Chance of a big question (4+ marks) here: 99%

Question types predicted here essay ×15

What each model expects

  • DeepSeek V4: An essay on the Korean War and Berlin blockade as early Cold War flashpoints, a clear gap.
  • Claude Fable 5: A crises-focused question on the 1961-1968 set anchors one option choice, while the alternative finally touches the renewal period (Afghanistan/Reagan) rather than another Gorbachev-or-détente rerun.
  • Gemini 3.1 Pro: A 25-mark essay focused on the Cuban Missile Crisis and its role in forcing a shift towards Détente.
  • GPT-5.6 Sol: The renewal phase—Afghanistan and Reagan—becomes the principal Cold War focus after origins, détente, arms and space races, and Gorbachev dominated 2024–2025.
  • Grok 4.6: Arms race, Gorbachev-alone, origins from 1945 and ‘why détente’ will be rested; Reagan/Afghanistan or a crises-and-relations judgement is due.
  • Claude Opus 5: The Cold War pair will draw one question from the 'Renewal and end' strand centred on Afghanistan or the collapse of communism rather than on Gorbachev, who was set in 2019 and again in 2025.
Sect III Opt C: Conflict in Europe 1935–1945 99% likely

Chance of a big question (4+ marks) here: 99%

Question types predicted here essay ×15

What each model expects

  • DeepSeek V4: An essay evaluating the role of appeasement and the Nazi–Soviet Pact in causing war, a survey gap.
  • Claude Fable 5: A civilians question naming the Holocaust in the occupied territories (first time as sole focus) is the sharpest call; the pairing question revisits either appeasement/collective security or North Africa.
  • Gemini 3.1 Pro: A 25-mark essay focused exclusively on the civilian experience in occupied territories, avoiding early military turning points tested in 2024 and 2025.
  • GPT-5.6 Sol: North Africa returns as a named campaign, paired with a civilians alternative centred more specifically on the Holocaust in occupied territories.
  • Grok 4.6: Origins to 1939, the Russian campaign and D-Day as turning point will be rested; North Africa is the seven-year cycle gap.
  • Claude Opus 5: With the early German advances and D-Day both set in 2025, 2026 moves back to the survey (collective security and appeasement) or to the war's end — Nuremberg and final defeat.
Sect III Opt B: Conflict in the Pacific 1937–1951 99% likely may rest

Chance of a big question (4+ marks) here: 98%

Question types predicted here essay ×10

What each model expects

  • DeepSeek V4: An essay on the fall of Singapore and the Dutch East Indies, a specific territory gap.
  • Claude Fable 5: The occupation-of-Japan/War Crimes Tribunals strand returns after five-plus years, most likely paired with a Midway/Guadalcanal turning-points judgement.
  • GPT-5.6 Sol: One Pacific alternative moves beyond the heavily tested combat years to Allied occupation, War Crimes Tribunals and the status of Hirohito.
  • Claude Opus 5: With Japanese occupation of South-East Asia and the 1941–42 advance both set in 2025, the 2026 pair shifts to the end of the conflict — occupation of Japan or the war crimes tribunals.
Sect II Opt G: USA 1919–1941 99% likely

Chance of a big question (4+ marks) here: 99%

Question types predicted here essay ×15

What each model expects

  • DeepSeek V4: An essay on the long-term causes of the Great Depression and reactions to the Crash, a survey gap.
  • Claude Fable 5: Consumerism/entertainment anchors one question after its longest-ever rest, paired with a New Deal (FDR-era) success evaluation following 2025's Hoover-only focus.
  • Gemini 3.1 Pro: A 25-mark essay evaluating the influence of conservative social movements in the 1920s, steering away from economic policy.
  • GPT-5.6 Sol: One USA alternative returns to the Depression’s differentiated impact, while the other targets mobilisation and war production to 1941.
  • Grok 4.6: Foreign policy and Depression/New Deal/Hoover will be rested; the live 25-marker is consumerism or racism as a social-change judgement.
  • Claude Opus 5: One USA option will name a specific group or force affecting society — consumerism, or women/farmers/African Americans in the Depression — following the 2022 and 2025 habit of naming the force in the question and inviting rebuttal.
Sect II Opt F: Russia and the Soviet Union 1917–1941 99% likely

Chance of a big question (4+ marks) here: 99%

Question types predicted here essay ×15

What each model expects

  • DeepSeek V4: An essay on the Civil War and NEP’s role in Bolshevik consolidation, a survey gap.
  • Claude Fable 5: One question centres on Stalin's social and cultural transformation (due since 2022); its pair is foreign policy's ideology-versus-pragmatism or the power struggle, with at least one stem in the 2025 'Evaluate the accuracy of this statement' form.
  • Gemini 3.1 Pro: A 25-mark essay focused on the social and cultural transformation under Stalin, stepping away from the economic policies tested in 2023 and 2025.
  • GPT-5.6 Sol: The paper pivots from consecutive economic, foreign-policy and Party-growth questions to repression and early Bolshevik consolidation.
  • Grok 4.6: One 25-marker will shift off power struggle, foreign policy and economic ideology onto social/cultural transformation or the nature of Stalinist control.
  • Claude Opus 5: One of the two Russia options will target Stalin's social and cultural transformation — the syllabus row rested since 2022 while economic and political transformation were both set in 2025.
Sect II Opt B: China 1927–1949 99% likely may rest

Chance of a big question (4+ marks) here: 99%

Question types predicted here essay ×12

What each model expects

  • DeepSeek V4: An essay evaluating the GMD’s achievements and limitations in the Nanjing decade, a clear syllabus gap.
  • Gemini 3.1 Pro: An essay question focused on evaluating the strategic blunders of the Guomindang versus the political success of the CCP.
  • GPT-5.6 Sol: The Nationalist decade and Yan’an replace the heavily examined Long March, Maoism and final Communist victory as the two China alternatives.
  • Grok 4.6: Long March, Maoism and ‘why the CCP won’ will be rested; Yan’an or the Nationalist decade 1927–1937 is the overdue 25-marker.
  • Claude Opus 5: China will set one question requiring the Maoism/Mao distinction and one on the Nationalists, continuing NESA's two-year insistence that candidates treat Maoism as an ideology in its own right.
Sect II Opt A: Australia 1918–1949 99% likely may rest

Chance of a big question (4+ marks) here: 98%

Question types predicted here essay ×7

What each model expects

  • DeepSeek V4: An essay on the 1949 election policies, requiring evaluation of their role in the Coalition victory against other factors like communism and postwar reconstruction.
  • Claude Fable 5: The Depression-and-political-responses row anchors one question after a six-year rest; the pair is communism's role in the 1949 election or a dual-element postwar reconstruction stem.
  • GPT-5.6 Sol: One Australia alternative will return to the Great Depression and political responses after three consecutive papers centred elsewhere.
Sect IV Opt B: Cultural Revolution to Tiananmen Square 1966–1989 99% likely may rest

Chance of a big question (4+ marks) here: 98%

Question types predicted here short answer ×3 extended response ×2 stimulus based ×1

What each model expects

  • DeepSeek V4: A 5-mark on Sino-Soviet relations, an 8-mark on education/culture impact with a photo, and a 12-mark on the protests’ impact on CCP leadership.
  • Grok 4.6: Deng’s reforms and Tiananmen’s world standing will be rested; Zhou Enlai’s death and Mao’s aims or competing Marxisms return.
Sect IV Opt F: Apartheid in South Africa 1960–1994 99% likely may rest

Chance of a big question (4+ marks) here: 99%

Question types predicted here short answer ×6 extended response ×6 stimulus based ×2

What each model expects

  • DeepSeek V4: A 5-mark on apartheid ideology and rural impact, an 8-mark on Bantustans with a map, and a 12-mark on internal political/economic factors.
  • Claude Fable 5: Sharpeville anchors a part (a)/(b) after its seven-year rest, and part (c) evaluates Mandela's leadership against other factors in ending apartheid, with the source in part (b) per the 2022-2025 alternation.
  • Gemini 3.1 Pro: The 8-mark question will focus explicitly on the Bantustans as a geographic and economic mechanism of control.
  • Grok 4.6: International responses, economic collapse and Mandela’s imprisonment will be rested; Soweto/rural Bantustans plus the transition to democracy are due.
  • Claude Opus 5: Apartheid's 12-mark part (c) will target the political and international factors in the collapse of apartheid, completing the economic-and-social pairing set in 2025.
Sect II Opt E: Japan 1904–1937 98% likely may rest

Chance of a big question (4+ marks) here: 100%

Question types predicted here essay ×12

What each model expects

  • Claude Fable 5: One question names Hirohito or the army's political divisions in the rise of militarism (rested since 2020); its pair evaluates a 1920s-democracy or tradition-versus-modernisation statement rather than repeating 2024-25's Depression and zaibatsu focuses.
  • Gemini 3.1 Pro: An essay on the economic catalysts (Great Depression, zaibatsu influence) that destabilised Japan's democracy in the 1930s.
  • GPT-5.6 Sol: Hirohito or internal army divisions will become the named factor after three years dominated by party politics, foreign policy, the Depression, militarism and the zaibatsu.
  • Grok 4.6: Hirohito or 1920s limited democracy returns; Great Depression, zaibatsu-as-challenge, and ‘was militarism why democracy failed’ will be rested.
  • Claude Opus 5: Japan's 2026 pair will avoid re-setting the zaibatsu and the collapse of party politics — used in 2023 and 2025 — in favour of social tensions in the 1930s or Hirohito's role.
Sect III Opt A: Conflict in Indochina 1954–1979 98% likely may rest

Chance of a big question (4+ marks) here: 100%

Question types predicted here essay ×13

What each model expects

  • Claude Fable 5: A dual-element strategies-and-tactics effectiveness question (NVA/NLF versus USA) appears — the one syllabus row never directly examined — paired with a Pol Pot or communist-victory judgement question.
  • Gemini 3.1 Pro: A 25-mark essay focused on the strategies and tactics of the North Vietnamese Army and Viet Cong.
  • GPT-5.6 Sol: After the 2025 Cambodia, Laos and civilian questions, the alternatives return to pre-1964 escalation and comparative military effectiveness.
  • Grok 4.6: Anti-war movements, civilians in Vietnam, and the spread into Cambodia/Laos will be rested; the live question is Communist victory or NVA/NLF tactics.
  • Claude Opus 5: With civilians in North and South Vietnam and the spread to Cambodia and Laos both set in 2025, 2026 turns to the reasons for the Communist victory and to the effectiveness of the belligerents' strategies.
Sect II Opt H: Iran 1945–1989 98% likely may rest

Chance of a big question (4+ marks) here: 0%

Sect II Opt D: Indonesia 1945–2005 98% likely may rest

Chance of a big question (4+ marks) here: 100%

Question types predicted here essay ×2

What each model expects

  • Gemini 3.1 Pro: A 25-mark question assessing the short- and long-term consequences of the 1965 coup and the violent transition to the New Order.
Sect II Opt C: India 1942–1984 98% likely may rest

Chance of a big question (4+ marks) here: 0%

Sect III Opt E: Conflict in the Gulf 1980–2011 98% likely may rest

Chance of a big question (4+ marks) here: 0%

Sect III Opt F: The Arab–Israeli Conflict 1948–1996 98% likely may rest

Chance of a big question (4+ marks) here: 0%

Sect IV Opt A: Pro-democracy movement in Burma 1945–2010 98% likely may rest

Chance of a big question (4+ marks) here: 0%

Sect IV Opt D: The Changing World Order 1945–2011 98% likely may rest

Chance of a big question (4+ marks) here: 0%

Sect IV Opt E: The Nuclear Age 1945–2011 98% likely may rest

Chance of a big question (4+ marks) here: 0%

How likely each topic is to appear. Open a topic for the question types to practise there.

Practise with the paper

Modern History practice paper

100 marks · 24 questions

Every question is traceable to the consensus prediction behind it — open the web version and each question carries a “why this question” link into the evidence. All questions are original Intuition compositions in NESA style.

Read it on the web Download PDF Solutions (PDF)

Intu AI

One paper isn't enough? Generate more

Intu AI builds unlimited practice questions for Modern History in these styles, marks your working, and explains what you missed — aligned to your syllabus.

Practise with Intu AI →

Share & save

Practice paper (PDF)

Every style card and paper question has its own link — hover any card and use its copy-link icon to share exactly the thing you mean. Printing this page gives a clean copy too.

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.