HSC 2026 published Aug 2026

Biology 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 Biology paper, marking guideline and marking-centre feedback report since 2019 — the same method we backtested against the real 2025 papers in Chemistry and Maths Extension 1 before publishing. These are styles to prepare for, not guarantees: the backtest showed examiners keep the skill and twist the format, so practise the skill chain, not a memorised question. One thing 2026 is not: the last year of this syllabus. The current Biology course runs through to the 2027 HSC, so expect normal rotation, not a farewell everything-paper.

Section II · 3–5 marks 6 of 6 models expect this

The pedigree — justify it, don't just name it

The annual fixture: a multi-generation pedigree for an unfamiliar disorder; students state the mode of inheritance (autosomal recessive most due after 2025's run, sex-linked the live...

Try it → Q24 Why we expect this

Section II · 5–7 marks 6 of 6 models expect this

The epidemiology evaluation — validity, reliability, then a verdict

A described cohort or case-control study (sample size, matching, duration, control group, confounders, self-reporting) linking an exposure to a non-infectious disease; students evaluate...

Try it → Q34 Why we expect this

Section II · 3–6 marks 6 of 6 models expect this

The transgenic-organism process

Describe the full production of a named transgenic organism for a new agricultural or medical target: isolate the gene with restriction enzymes (sticky ends), insert into a plasmid vector...

Try it → Q35 Why we expect this

On this page

1. The pedigree — justify it, don't just name it
2. The epidemiology evaluation — validity, reliability, then a verdict
3. The codon-chart mutation question
4. Blood glucose negative feedback — the due data question
5. The transgenic-organism process
Where to spend study time
Practice paper
Check our working

The big five (have these cold)

1. The pedigree — justify it, don't just name it

What it looks like: 3–5 marks: a multi-generation pedigree for an unfamiliar disorder; state the mode of inheritance and justify it by naming specific numbered individuals, then a keyed Punnett square for one couple and an offspring probability. After 2025's autosomal-dominant run, autosomal recessive is most due — but be ready for sex-linked.
Why we expect it: it has appeared in five of the last seven papers and all six models on our panel predicted it — the strongest consensus on the board.
The traps markers flag (2021, 2022, 2025): justifying with generic rules instead of citing individuals ("III-2 is affected but her parents are not, so the allele is recessive"); omitting the key; choosing allele letters you can't tell apart (S and s in your handwriting); giving the genotypic ratio when the phenotypic ratio was asked; writing "somatic" when you mean "autosomal".
Practise: one pedigree a week, and always write the exclusion argument — why it isn't the other three modes.

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

2. The epidemiology evaluation — validity, reliability, then a verdict

What it looks like: 5–7 marks late in the paper: a described study (cohort size, duration, matching, self-reporting) linking an exposure to a non-infectious disease; evaluate the method and give an explicit judgement.
Why we expect it: it has run in five of the last six papers. In 2025 the examiners flipped it to "design a study" — the panel expects reversion to evaluate form, but prepare both directions.
The traps (2021, 2023, 2025): treating validity, reliability and accuracy as synonyms; evaluating without ever delivering the judgement the verb demands; ignoring the specific features in the stimulus (sample size, confounders, missing control) in favour of a memorised checklist.
Practise: end every response with a one-sentence verdict that weighs a strength against a weakness of the actual study described.

Try it → Q34 6 of 6 models expect this — see the evidence

3. The codon-chart mutation question

What it looks like: 3–6 marks: a normal and a mutated DNA template strand plus a codon chart; transcribe to mRNA, translate both, classify the mutation (silent / missense / nonsense / frameshift) and explain the consequence for protein folding and function.
Why we expect it: it has run three years straight (2023, 2024, 2025) and five of six models predict a fourth. One model is contrarian and expects the chart itself to rest — but the mutation-to-protein reasoning chain appears either way.
The traps (2023, 2024, 2025): transcribing wrongly before touching the chart (everything downstream dies); forgetting U replaces T; stopping at the amino-acid change without linking it to shape and function; not seeing a frameshift reach across every downstream codon.
Practise: the full chain — DNA → mRNA → amino acids → shape → function — until it is automatic in both the normal and mutant lane.

Try it → Q26 5 of 6 models expect this — see the evidence

4. Blood glucose negative feedback — the due data question

What it looks like: 4–7 marks: time-series graphs of blood glucose, insulin and glucagon after a meal; name receptor, control centre and effectors — pancreatic alpha and beta cells, the liver, glycogen — and explain both hormonal pathways citing turning points in the actual traces.
Why we expect it: glucose control has not carried a substantial written question since 2020 (9 marks), while recent homeostasis questions went to temperature and water — the clearest coverage hole in Module 8. Five models call its return; one adds a "how would an untreated Type 2 trace differ?" twist.
The traps (2020, 2021): naming insulin but not glucagon; swapping alpha and beta cells; describing the graphs without integrating all of them; never quoting a number off the axes.

Try it → Q28 5 of 6 models expect this — see the evidence

5. The transgenic-organism process

What it looks like: 3–6 marks: describe producing a named transgenic organism — cut the gene out with a restriction enzyme (sticky ends), insert into a plasmid vector with DNA ligase, amplify in bacteria, deliver into germ-line cells so the trait is heritable.
Why we expect it: the 2021 salmon / 2023 insulin / 2025 mosquito lineage; all six models predict it continues, most likely at fewer marks after 2025's 11-mark treatment, on a fresh organism.
The traps (2023, 2025): "place the gene into" instead of the enzyme steps; calling ligase a cutting enzyme; forgetting germ-line delivery — a gene in liver cells doesn't make a transgenic lineage.

Try it → Q35 6 of 6 models expect this — see the evidence

Also on the radar

Gene pool: drift vs gene flow vs selection — allele-frequency data for an isolated population; the single most repeated marker complaint since 2020 is students swapping gene flow and genetic drift. Population size is the tell: drift needs small.
DNA replication in written form — untouched as a big written item since 2022; expect semi-conservative strand-tracking, or a comparison with transcription on shared criteria (enzymes, template, product, uracil) — markers punish two separate descriptions posing as a comparison.
Somatic vs germ-line mutation — flagged in four feedback reports; only germ-line mutations reach offspring. Say which cells, not just which word.
Vaccination-program data — incidence before and after a program: quote the data, explain herd immunity, and note what the data cannot establish (correlation is not causation).
Biotechnology vs biodiversity — a table of technologies (selective breeding, artificial insemination, cloning, transgenics); judge each row's effect on the gene pool, not on the organism. The odd-year strand that skipped 2025.
Non-Mendelian crosses — codominance, incomplete dominance, blood-group multiple alleles; keyed Punnett square, phenotypic ratio.
Pathogen ID plus a named adaptation — the small-mark opener: classify by structure or a key, then an entry or transmission adaptation of a named pathogen (not the disease) — and don't hand in an immune-evasion story.
Kidney dialysis — the technologies rotation points here after vision (2025) and hearing (2024): urea diffuses down its gradient across a semi-permeable membrane. "Dialysis cleans the blood" scores nothing.
Non-disjunction with a karyotype — three models call a written return after 2025 kept it multiple-choice: name the abnormality, then meiosis I vs II and fertilisation.
The mandated agar-plate practical — a full safe design (IV, DV, uninoculated control, repetition, incubate and never reopen) last set in full in 2022.

Where to spend less time

The panel's rested calls: reproduction carried 10+ marks in 2025 (fertilisation data, fungal reproduction) and three of six models formally rest it — expect multiple choice and small marks, not another anchor. Same story for the adaptive immunity cascade: after 9–11 mark treatments in 2024 and 2025, another extended B-and-T-cell sequence is the panel's least likely big question (the lowest substantial-question probability of all eleven topics) — the live risk is instead a small innate-immunity or passive/active item. Know both — everything is examinable — but don't build your revision around them.

The honest fine print

We haven't backtested Biology specifically; we backtested the method, on the real 2025 papers in Chemistry and Maths Extension 1. There, every topic we rated ≥90% appeared (37/37 across both subjects), and roughly half of the specific question predictions recognisably appeared — the misses clustered where examiners twisted a format, moved a question into multiple choice, or broke a streak. Biology's marking-feedback record suggests the same pattern will hold. So treat the probabilities above as strong guidance about skills and stimulus types, not a script: master the skill chains — justify from the pedigree, transcribe before you translate, name the actual cells and ions and individuals — and you're covered whichever way the format twists.

Want to check our working? every call above, with each model's own prediction

Pedigree: identify and justify the mode of inheritance, with keyed Punnett square 6 of 6 models expect this

bio-q1-pedigree-mode-of-inheritance Section II 3–5 marks mid range consensus 0.71

The annual fixture: a multi-generation pedigree for an unfamiliar disorder; students state the mode of inheritance (autosomal recessive most due after 2025's run, sex-linked the live alternative) and justify it by naming specific numbered individuals whose children exclude the alternatives, then construct a keyed Punnett square for one couple and give an offspring probability or phenotypic ratio. Appeared 2019, 2021, 2023, 2024, 2025.

Each model's own prediction

Claude Fable 5: autosomal recessive due after 2025, key + individuals in the marks (p 0.80)
Claude Opus 5: three-generation pedigree, exclude alternatives via named individuals (p 0.72)
GPT-5.6 Sol: sex-linked/multiple-allele pedigree with obligate carriers and probability (p 0.56)
Gemini 3.1 Pro: sex-linked Punnett from a pedigree in an unfamiliar organism (p 0.85)
Grok 4.6: justify autosomal recessive, genotype of a numbered individual (p 0.72)
DeepSeek V4: pedigree genotypes plus probability of an affected child (p 0.60)

Marker-feedback lineage: Marking feedback 2021 Q24(a), 2022, 2023 Q25(a), 2024 Q20, 2025 Q31(a)

In the practice paper: Q24

Epidemiological-study evaluation: validity vs reliability with an explicit judgement 6 of 6 models expect this

bio-q2-epidemiology-study-evaluation Section II 5–7 marks discriminator consensus 0.67

A described cohort or case-control study (sample size, matching, duration, control group, confounders, self-reporting) linking an exposure to a non-infectious disease; students evaluate the method — critiquing reliability distinctly from validity — and deliver an explicit judgement on whether the evidence supports the link. The annual fixture (2020, 2021, 2022, 2023, 2025), returning to evaluate form after 2025's design-a-study variant.

Each model's own prediction

Claude Fable 5: returns to evaluate form after 2025's design variant (p 0.75)
Claude Opus 5: cohort study with explicit judgement demanded (p 0.65)
GPT-5.6 Sol: 6-8 mark capstone incl. incidence/prevalence and causation judgement (p 0.62)
Gemini 3.1 Pro: critique reliability distinctly from validity (p 0.85)
Grok 4.6: named pollutant, ~500 people, 12 months, three cities (p 0.58)
DeepSeek V4: evaluate method and results with provided data (p 0.55)

Marker-feedback lineage: Marking feedback 2021 Q31, 2022 Q31(a), 2023 Q27, 2025 Q32

In the practice paper: Q34

Codon-chart mutation: transcribe, classify, link to protein structure and function 5 of 6 models expect this

bio-q3-codon-chart-mutation Section II 3–6 marks discriminator consensus 0.64

A template DNA strand (normal and mutant) with a supplied codon chart: students transcribe to mRNA, translate both sequences, classify the mutation (silent/missense/nonsense substitution vs frameshift), and explain the consequence for polypeptide folding and protein function. The 2023 Q25(b) / 2024 Q28(b) / 2025 Q31(b) lineage continued.

Each model's own prediction

Claude Fable 5: mutant amino acid sequence or mutagen diagram, classify then explain (p 0.70)
Claude Opus 5: back-translate, name substitution/frameshift, consequence for folding (p 0.60)
GPT-5.6 Sol: regulatory vs coding gene, silent/missense/nonsense (p 0.53)
Gemini 3.1 Pro: frameshift across two codons, multiple amino acids altered (p 0.70)
DeepSeek V4: 6-mark coding-strand point mutation via chart to function (p 0.65)

Marker-feedback lineage: Marking feedback 2023 Q25(b), 2024 Q28(b), 2025 Q31(b)

A note on this agreement

Grok is the deliberate contrarian: after the 2023-2025 three-year streak it predicts the 2026 variant drops the full chart calculation (p 0.42, counted as disagreeing). All surface details grepped clean against the corpus.

In the practice paper: Q26

Somatic vs germ-line mutation: heritability and consequence 5 of 6 models expect this

bio-q4-somatic-vs-germline Section II 2–5 marks mid range consensus 0.59

A named mutagen (UV, X-rays or a chemical) or a cell-lineage diagram marking where mutations occur; students distinguish somatic from germ-line mutations and justify which can be inherited by offspring, contrasting consequence for the individual versus the population. Echoes 2021 Q24(b) and 2025 Q33(c); the panel's most-repeated marker-feedback theme after gene flow/drift.

Each model's own prediction

Claude Fable 5: cell-differentiation lineage diagram, which offspring carry each mutation (p 0.45)
Claude Opus 5: named mutagen, molecular damage, individual vs offspring (p 0.50)
GPT-5.6 Sol: MC placing mutations in coding DNA/promoter/intron/gamete (p 0.49)
Gemini 3.1 Pro: electromagnetic radiation, germ-line vs somatic evolutionary consequence (p 0.80)
DeepSeek V4: outline the difference and the effect on offspring (p 0.70)

Marker-feedback lineage: Marking feedback 2019, 2021, 2024, 2025 (somatic/germ-line confusion flagged four times)

In the practice paper: Q22

Vaccination/prevention program judged against incidence data (herd immunity) 5 of 6 models expect this

bio-q5-vaccination-program-analysis Section II 4–7 marks discriminator consensus 0.60

Disease-incidence or antibody data before and after a named prevention measure (vaccination program the favourite); students quantify the trends, link each procedure to interrupted transmission, explain herd immunity for the unvaccinated minority, and judge effectiveness — including what the data cannot establish (correlation vs causation, time lags). The 2021 Q30 / 2023 Q30 mould.

Each model's own prediction

Claude Fable 5: antibody-schedule or case/death data, herd immunity (p 0.55)
Claude Opus 5: assessment distinguishing active from passive, what data cannot show (p 0.55)
GPT-5.6 Sol: vector-borne measures compared, delayed effects and mobility (p 0.57)
Gemini 3.1 Pro: ten-year incidence graph, named prevention method (p 0.80)
DeepSeek V4: incidence before/after program plus herd immunity (p 0.55)
Grok 4.6: contrarian - campaign-effectiveness logic as MC only (p 0.38, not counted)

Marker-feedback lineage: Marking feedback 2021 Q30, 2023 Q30; 2025 MC10 correlation trap

In the practice paper: Q32

Transgenic organism production: restriction enzymes, ligase, plasmid, germ-line delivery 6 of 6 models expect this

bio-q6-transgenic-production-process Section II 3–6 marks mid range consensus 0.57

Describe the full production of a named transgenic organism for a new agricultural or medical target: isolate the gene with restriction enzymes (sticky ends), insert into a plasmid vector with DNA ligase, amplify in bacteria, deliver into germ-line cells so the trait is heritable. The 2021 salmon / 2023 insulin / 2025 mosquito lineage, likely at reduced marks after 2025's 11-mark treatment.

Each model's own prediction

Claude Fable 5: novel target at reduced marks after 2025 (p 0.50)
Claude Opus 5: full step sequence ending in heritable germ-line transformation (p 0.62)
GPT-5.6 Sol: via gene-cloning vs whole-organism-cloning contrast (p 0.39)
Gemini 3.1 Pro: sequence enzymes/plasmid/ligase, avoid breeding descriptions (p 0.80)
Grok 4.6: medical host (insulin/growth hormone) as complement to 2025's mosquito (p 0.52)
DeepSeek V4: recombinant insulin process (p 0.60)

Marker-feedback lineage: Marking feedback 2021 Q33(a), 2023 Q31, 2025 Q30(a)

In the practice paper: Q35

Blood-glucose negative feedback from insulin/glucagon traces 5 of 6 models expect this

bio-q7-glucose-negative-feedback Section II 4–7 marks discriminator consensus 0.57

Time-series plots of blood glucose with insulin and glucagon after a carbohydrate meal (or exercise): students name stimulus, receptor, control centre and effectors — pancreas alpha/beta cells, liver, glycogen — explain both hormonal pathways citing turning points in the traces, and demonstrate negative feedback. Last written substantially in 2020 Q31 (9 marks): the clearest coverage hole in Module 8. Grok adds an untreated Type 2 contrast as the twist.

Each model's own prediction

Claude Fable 5: insulin/glucagon traces or glucose-tolerance curves (p 0.50)
Claude Opus 5: antagonistic-hormone traces, calcium/PTH the alternative analyte (p 0.55)
GPT-5.6 Sol: three linked time-series, turning points demanded (p 0.64)
Grok 4.6: five-hour post-meal graphs plus Type 2 trace contrast (p 0.62)
DeepSeek V4: predict the effect of a beta-cell defect (p 0.55)

Marker-feedback lineage: Marking feedback 2020 Q31(a), 2021 Q32; four-year written gap

In the practice paper: Q28

Gene-pool processes from population data: drift vs gene flow vs selection 5 of 6 models expect this

bio-q8-gene-pool-processes Section II 4–7 marks discriminator consensus 0.57

Allele-frequency tables or graphs for isolated versus connected populations (island/bottleneck scenarios): students calculate or read frequencies and explain which process — mutation, gene flow, genetic drift, natural selection — produced the change, justifying why population size matters. The 2022 Q32 / 2023 Q32 / 2025 Q34 lineage; gene flow vs drift is the single most repeated marker complaint.

Each model's own prediction

Claude Fable 5: isolated vs connected populations, all four processes (p 0.65)
Claude Opus 5: bottleneck/founder drift distinguished from flow and selection (p 0.55)
GPT-5.6 Sol: SNP profiles, conservation breeding, small-sample limitation (p 0.44)
Gemini 3.1 Pro: calculate frequencies, flow vs drift verdict (p 0.70)
Grok 4.6: island-disaster MC with drift as key (p 0.50)

Marker-feedback lineage: Marking feedback 2020 Q29, 2021 Q26, 2022 Q32, 2023 Q32, 2025 Q34

In the practice paper: Q33

DNA replication returns in written form (semi-conservative; compare with transcription) 6 of 6 models expect this

bio-q9-dna-replication-written-return Section II 3–6 marks mid range consensus 0.57

DNA replication's first substantial written outing since 2022 Q28: a labelled-strand tracking task identifying the semi-conservative model, a structured comparison of replication with transcription on shared criteria (enzymes, template, product, location, uracil), or the prokaryote/eukaryote DNA contrast (circular vs linear, histones) — possibly seeded by a Chargaff-style base-proportion step extending 2025 MC19.

Each model's own prediction

Claude Fable 5: semi-conservative or prok/euk compare, Chargaff extension (p 0.45)
Claude Opus 5: compare transcription with replication on shared criteria (p 0.45)
GPT-5.6 Sol: labelled strands tracked through two cycles (p 0.61)
Gemini 3.1 Pro: table comparing replication and transcription (p 0.80)
Grok 4.6: prok/euk DNA structural differences (p 0.38); mRNA base-pairing MC (p 0.70)
DeepSeek V4: describe replication with enzymes, semi-conservative (p 0.70)

Marker-feedback lineage: Marking feedback 2022 Q28, 2024 Q30(a); 2021-2024 replication/synthesis confusion

In the practice paper: Q25

Pathogen classification plus a named adaptation for entry or transmission 6 of 6 models expect this

bio-q10-pathogen-classification-adaptation Section II 2–4 marks routine consensus 0.56

The small-mark Section II opener strand: classify pathogens by structural features or a dichotomous key (virus vs bacterium vs prion vs fungus), then outline one adaptation of a NAMED pathogen that facilitates entry into or transmission between hosts — not survival or immune evasion. Recurs 2019 Q31, 2021 Q21, 2022 Q21, 2023 Q28, 2025 Q22(b).

Each model's own prediction

Claude Fable 5: classification key then named-pathogen adaptation (p 0.65)
Claude Opus 5: transmission not survival - cholera toxin example (p 0.40)
GPT-5.6 Sol: fungal pathogen of an Australian plant, entry vs response (p 0.48)
Gemini 3.1 Pro: biofilm/antigenic variation in a clinical scenario (p 0.80)
Grok 4.6: dichotomous-key completion plus H. pylori-style adaptation (p 0.50)
DeepSeek V4: virus vs bacterium entry adaptations compared (p 0.50)

Marker-feedback lineage: Marking feedback 2019 Q31(a), 2022 Q21(b), 2023 Q28, 2025 Q22(b)

In the practice paper: Q21

Biotechnologies evaluated against biodiversity at gene-pool and species level 5 of 6 models expect this

bio-q11-biotech-biodiversity-evaluation Section II 4–7 marks discriminator consensus 0.55

A table or dataset of technologies (selective breeding, artificial insemination, cloning, transgenics such as Bt cotton) applied to a crop or livestock species; students judge whether each increases, maintains or decreases genetic and species biodiversity, linking the technology to the gene pool rather than to the organism. The odd-year strand (2019 Q24, 2021 Q33(c), 2023 Q34) that skipped 2025.

Each model's own prediction

Claude Fable 5: table of technologies, judge each row (p 0.55)
Claude Opus 5: net effect at gene-pool and species level (p 0.48)
Gemini 3.1 Pro: named agricultural biotech vs regional biodiversity (p 0.70)
Grok 4.6: AI + transgenic crop on cattle/crop diversity (p 0.48)
DeepSeek V4: Bt cotton impact with data (p 0.55)

Marker-feedback lineage: Marking feedback 2019 Q24, 2021 Q33(c), 2023 Q34, 2024

In the practice paper: Q35

Non-Mendelian cross: codominance / incomplete dominance / multiple alleles 4 of 6 models expect this

bio-q12-non-mendelian-cross Section II 3–4 marks mid range consensus 0.55

A cross in an unfamiliar organism whose F1/F2 data reveal codominance, incomplete dominance or multiple alleles (blood-group style IA, IB, i): students assign genotypes, construct a keyed Punnett square and state the phenotypic — not genotypic — ratio. Last written 2022 Q22 (eggplant); MC-only in 2025.

Each model's own prediction

Claude Fable 5: 1:2:1 or blended/spotted phenotypes, keyed square (p 0.45)
Claude Opus 5: deduce the pattern from stated parental phenotypes (p 0.62)
Grok 4.6: heterozygous cross, blood-group alleles the favourite (p 0.58)
DeepSeek V4: compare codominance and incomplete dominance with examples (p 0.55)

Marker-feedback lineage: Marking feedback 2019 Q30, 2021 Q22, 2022 Q22, 2024 Q28(a)

In the practice paper: Q23

Kidney function and dialysis: the rotation's due organ technology 5 of 6 models expect this

bio-q13-kidney-dialysis Section II 3–6 marks mid range consensus 0.53

The Module 8 technologies written question rotates to the kidney after vision (2025) and hearing (2024): clinical data (declining GFR, rising blood urea) establish loss of function; students explain how haemodialysis compensates — urea diffusing down its concentration gradient across a semi-permeable membrane into dialysate, blood cells and useful solutes retained. Unwritten since 2020 Q24.

Each model's own prediction

Claude Fable 5: GFR-style or blood-composition data, counter-current dialysate (p 0.55)
Claude Opus 5: declining GFR/urea data (p 0.40)
GPT-5.6 Sol: graph the trend, decide when dialysis is required (p 0.60)
Grok 4.6: concentration-arrow diagram, glucose retained (p 0.58)
DeepSeek V4: explain how dialysis compensates (p 0.50)

Marker-feedback lineage: Marking feedback 2020 Q24(c); 2022 Q20 and 2025 Q9 kept it MC-only

In the practice paper: Q30

Hearing technologies discriminated by site of lesion 6 of 6 models expect this · one echo discounted

bio-q14-hearing-technologies Section II 1–5 marks mid range consensus 0.52

Match hearing aid, bone-conduction device and cochlear implant to the site of damage (blocked outer ear with intact cochlea → bone conduction; damaged cochlea → implant), explaining whether each amplifies sound, bypasses the outer/middle ear, or directly stimulates the auditory nerve — as MC (2019 MC6, 2023 MC1, 2025 MC15) or a written justification with outcome data (2024 Q35, 2021 Q25(c)).

Each model's own prediction

Claude Fable 5: site-of-damage MC (p 0.70); outcome-data evaluation (p 0.40)
Claude Opus 5: audiogram or described pathology, justify bone conduction (p 0.38)
GPT-5.6 Sol: audiogram + structural diagnosis for two patients (p 0.41)
Gemini 3.1 Pro: audiogram-based recommendation (p 0.70)
Grok 4.6: audiogram or lesion stem, cochlea intact (p 0.40)
DeepSeek V4: cochlear-implant effectiveness by age from graph data (p 0.55)

Marker-feedback lineage: Marking feedback 2021 Q25(c), 2024 Q35

Why we discounted part of this agreement

Corpus echo: opus, gpt-5.6-sol, gemini-3.1-pro and grok all pose the stimulus as an "audiogram" — a term with ZERO occurrences in extract/biology.md (past papers use described pathologies and outcome tables). The audiogram framing is treated as one vote, not four; the device-discrimination substance is corpus-grounded and stands.

Per our corpus-echo rule, identical invented details count as one vote, not independent confirmation.

In the practice paper: Q18

Watch list — worth having ready (10)

Non-disjunction karyotype: identify the chromosomal abnormality and explain meiosis I/II non-disjunction plus fertilisation (gemini 0.75, grok 0.55, deepseek 0.60 - the strongest three-model cluster; written form due after 2025's MC-only Down syndrome)
Design the mandated agar-plate practical in full - IV/DV, uninoculated control, repetition, never reopening incubated plates (grok 0.55 as its contrarian bold call, gpt-5.6-sol 0.58, fable 0.50, deepseek 0.50, opus 0.42; full design last set 2022 Q26)
Innate/adaptive immunity kept small after 2024-25's 9-11 mark anchors: a Section I classification MC (fable 0.85) plus a 2-3 mark written on memory cells or passive immunity (deepseek 0.55, grok 0.50, opus 0.45)
Plant water balance homeostasis - ABA/guard cells, detection of internal state (grok 0.55, deepseek 0.50, fable 0.45, opus 0.45; gemini 0.75 as a xerophyte-adaptation variant)
Plot-and-extrapolate disease-incidence graphing, the even-year fixture absent 2025 (fable 0.60, opus 0.42)
Incidence vs prevalence reasoning - treatment prolongs life, prevalence rises (fable 0.60 MC, opus 0.45 written calculation, grok 0.45, deepseek 0.45)
Oxytocin positive-feedback contrast at birth (opus 0.45, grok 0.32 MC, fable as the twist inside a feedback flow-chart item 0.55)
Koch's postulates applied to an unfamiliar outbreak, absent 2025 (fable 0.50, deepseek 0.50, opus 0.45)
Plant responses to a named pathogen get their first substantial written item (opus bold call 0.30, fable bold call; MC-only 2023-24 - a genuine seven-year coverage gap)
Reproduction is the panel's likely-rested call after 2025 spent 10+ marks there (fable rests it at substantial level, gpt-5.6-sol 0.66, gemini 0.30 P(examined); MC and small marks only)

Where to spend your study time

How likely each topic is to appear this year.

Mod 5: Genetic variation, inheritance patterns, pop. genetics 95% likely

Chance of a big question (4+ marks) here: 79%

Question types predicted here short answer ×7 stimulus based ×5 extended response ×2 practical analysis ×1 multiple choice ×1

What each model expects

DeepSeek V4: A 5-mark pedigree/genotype probability question will be the discriminator, with a 3-mark comparison of inheritance types.
Claude Fable 5: A 3-4 mark pedigree question requiring named individuals and a keyed Punnett square appears mid-Section II, paired with a DNA-profile or electrophoresis interpretation part.
Gemini 3.1 Pro: A major pedigree or Punnett analysis in Section II demanding explicit calculation of phenotypic ratios.
GPT-5.6 Sol: The inheritance discriminator will use a pedigree requiring an explicit keyed genetic cross and a probability justified from family evidence.
Grok 4.6: Section II will include a 3–4 mark pedigree justification naming individuals, plus a separate 3-mark Punnett with a required key.
Claude Opus 5: A pedigree with lettered or numbered individuals will demand a justified mode of inheritance in the first half of Section II, worth 3–4 marks, with marks reserved for naming the individuals that exclude sex-linkage.

Mod 5: Cell replication, DNA structure and replication 95% likely

Chance of a big question (4+ marks) here: 75%

Question types predicted here stimulus based ×7 short answer ×6 extended response ×2 multiple choice ×1

What each model expects

DeepSeek V4: A 6-mark question will combine polypeptide synthesis with a mutation, requiring use of a codon chart and explanation of effect on protein structure.
Claude Fable 5: A 4-6 mark transcription/translation question in an unfamiliar protein context appears, with mRNA/tRNA roles as the mark-bearing detail, plus a mitosis/meiosis MC.
Gemini 3.1 Pro: A discriminator comparing the exact characteristics of DNA replication and transcription.
GPT-5.6 Sol: A 4–6 mark Section II DNA-replication model will require students to distinguish replication from transcription and justify semi-conservative products.
Grok 4.6: DNA replication versus transcription will not be re-asked after 2024 Q30; a faulty meiosis model or prok/euk DNA outline is the likelier written item.
Claude Opus 5: Polypeptide synthesis will be examined through a supplied diagram demanding the mRNA and tRNA roles in sequence, with the discriminating marks lying in codon-to-anticodon complementarity rather than recall of the steps.

Mod 6: Mutation and genetic change 94% likely

Chance of a big question (4+ marks) here: 76%

Question types predicted here short answer ×6 extended response ×4 stimulus based ×4 multiple choice ×2

What each model expects

DeepSeek V4: A 6-mark question on chromosomal mutation (non-disjunction) will discriminate, combined with a 3-mark question on somatic vs germ-line.
Claude Fable 5: A codon-chart translation-plus-mutation-classification item appears in Section II for a fourth consecutive year, worth 3-5 marks, with a gene-pool evaluation question elsewhere in the paper.
Gemini 3.1 Pro: A mid-range question requiring the contrast of evolutionary impacts between somatic and germ-line mutations.
GPT-5.6 Sol: A mutation question will require a complete DNA-to-protein consequence chain rather than merely naming the mutation.
Grok 4.6: Expect a 4–5 mark non-disjunction/karyotype explain, not another codon-chart amino-acid hunt.
Claude Opus 5: Mutation will again be the heaviest Module 6 strand, anchored by a codon-chart decoding item and a separate gene-pool question where drift and gene flow must be told apart using supplied allele-frequency data.

Mod 8: Homeostasis 92% likely

Chance of a big question (4+ marks) here: 73%

Question types predicted here short answer ×7 stimulus based ×6 extended response ×2 multiple choice ×1

What each model expects

DeepSeek V4: A 5-mark graph interpretation question on blood glucose feedback will be the discriminator, with a 3-mark adaptation question as mid-range.
Claude Fable 5: Glucose regulation returns as the substantial homeostasis item (4-6 marks with hormone-trace data) after a six-year written absence, alongside a 2-3 mark loop-completion or endotherm item.
Gemini 3.1 Pro: A stimulus-first interpretation of an unfamiliar negative feedback loop, following 2025's flowchart style.
GPT-5.6 Sol: Blood-glucose regulation returns as the principal homeostasis response, using linked glucose, insulin and glucagon graphs.
Grok 4.6: A 5–6 mark insulin–glucagon–glucose data question returns after a four-year gap, with plant water as a 3–4 mark partner item.
Claude Opus 5: Homeostasis returns as a substantial data-driven negative-feedback question worth 5–7 marks, the largest Module 8 physiology item since 2020, after being reduced to a trivial four-mark flow chart in 2025.

Mod 8: Causes of non-infectious disease, epidemiology 92% likely

Chance of a big question (4+ marks) here: 73%

Question types predicted here extended response ×5 short answer ×4 practical analysis ×3 multiple choice ×3 stimulus based ×1

What each model expects

DeepSeek V4: A 6-mark evaluation of an epidemiological study will be the discriminator, with a 3-mark calculation of incidence/prevalence.
Claude Fable 5: An evaluate-the-epidemiological-study question worth 5-7 marks sits in the final third of Section II, and the plot/best-fit/extrapolate graphing item returns after skipping 2025.
Gemini 3.1 Pro: A 5-7 mark extended response evaluating the reliability and validity of an epidemiological study design.
GPT-5.6 Sol: A 6–8 mark epidemiological-study evaluation will require both a methodological judgement and an incidence/prevalence calculation or distinction.
Grok 4.6: After a design-a-study year, 2026 reverts to a 6–7 mark evaluate-the-method epidemiology item with an explicit judgement required.
Claude Opus 5: An epidemiology question worth 5–7 marks will sit late in Section II and require an explicit validity judgement rather than a description of the data, with marks lost for hedging.

Mod 7: Causes and transmission of infectious disease 91% likely

Chance of a big question (4+ marks) here: 66%

Question types predicted here short answer ×7 stimulus based ×4 practical analysis ×3 extended response ×1 multiple choice ×1

What each model expects

DeepSeek V4: A 5-mark question on Koch's postulates and experimental design will appear, with a 3-mark question on transmission modes.
Claude Fable 5: Koch's postulates return as a written application question after MC-only treatment in 2024, and a low-numbered 2-3 mark pathogen classification/adaptation item opens Section II.
Gemini 3.1 Pro: A stimulus-driven scenario requiring explanation of pathogen transmission adaptations rather than host responses.
GPT-5.6 Sol: A practical-analysis question will combine microbial testing with classification of a bacterial versus viral pathogen.
Grok 4.6: A 3-mark pathogen classification key returns, with a 2–3 mark named-pathogen entry adaptation beside it.
Claude Opus 5: Module 7 causes will be examined through short 2–4 mark applied items rather than one large question, with at least one requiring a named pathogen rather than a named disease.

Mod 7: Prevention, treatment, control incl. epidemiology 91% likely

Chance of a big question (4+ marks) here: 70%

Question types predicted here stimulus based ×6 short answer ×4 extended response ×3 practical analysis ×2 multiple choice ×1

What each model expects

DeepSeek V4: A 5-mark evaluation of vaccination data will appear, with a 3-mark question on prevention procedures.
Claude Fable 5: Prevention/control anchors a 5-7 mark data question in the back half of Section II, framed around an unfamiliar named outbreak with control procedures justified against the data.
Gemini 3.1 Pro: An evaluation of a prevention method's effectiveness explicitly tied to a graphical data set.
GPT-5.6 Sol: A 5–7 mark evaluation will compare several controls for a vector-borne disease using incidence or distribution data.
Grok 4.6: Section II will restore a 5-mark microbial investigation design after 2023–2025 only asked fragments of experimental method.
Claude Opus 5: A vaccination graph will carry a 4–7 mark assess or evaluate item in Section II after being confined to three multiple-choice questions in 2025, with marks for an explicit judgement.

Mod 6: Biotechnology and genetic technologies 90% likely

Chance of a big question (4+ marks) here: 69%

Question types predicted here extended response ×7 short answer ×5 stimulus based ×3 multiple choice ×1

What each model expects

DeepSeek V4: A 6-mark evaluation of a biotechnology's impact on biodiversity will be the discriminator, with a 3-mark process description.
Claude Fable 5: The 5-7 mark biotechnology evaluation targets biodiversity effects in agriculture (the strand that skipped 2025), while the transgenesis-steps question shrinks to 3-4 marks after 2025's Q30.
Gemini 3.1 Pro: A highly specific question demanding exact biochemical terminology for transgenic production (plasmids, ligases).
GPT-5.6 Sol: Biotechnology is reduced from the 2024–2025 extended responses to a compact CRISPR or cloning process question with one evaluated consequence.
Grok 4.6: No second consecutive 11-mark ‘evaluate genetic technologies’ finale; instead a 4–5 mark medical recombinant-DNA outline and/or a 5–7 mark biodiversity evaluate.
Claude Opus 5: A 3–5 mark 'describe the process' item on producing a transgenic organism will appear, with marks tied to naming restriction enzymes, ligase, a vector and germ-line delivery in correct sequence.

Mod 8: Technologies for hearing/vision/kidney disorders 86% likely

Chance of a big question (4+ marks) here: 57%

Question types predicted here stimulus based ×7 short answer ×6 multiple choice ×2 extended response ×1

What each model expects

DeepSeek V4: A 4-mark question on cochlear implants with data will appear, and a 5-mark question comparing technologies for a visual disorder.
Claude Fable 5: The organ rotation (hearing 2024, vision 2025) delivers a 4-6 mark kidney/dialysis question — the first substantial kidney item since 2020 — plus a hearing-technology MC.
Gemini 3.1 Pro: A 4-5 mark question comparing exactly how two different technologies correct a specific visual or auditory disorder.
GPT-5.6 Sol: Kidney failure and dialysis will replace the recent emphasis on vision and cochlear implants as the substantial technologies-and-disorders context.
Grok 4.6: A 3–4 mark dialysis mechanism question returns; vision stays at MC after 2025 LASIK.
Claude Opus 5: Loss of kidney function and dialysis reappears in Section II for the first time since 2020, worth 3–6 marks, since vision was covered in 2023 and 2025 and hearing in 2021 and 2024.

Mod 5: Reproduction and continuity of species 72% likely may rest

Chance of a big question (4+ marks) here: 51%

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

What each model expects

DeepSeek V4: A 3-mark comparison of reproduction in fungi and a 4-mark question on artificial pollination will be present.
Grok 4.6: Internal versus external fertilisation will not be re-asked after 2025 Q27; pregnancy-hormone table or implantation/birth MC fills the reproduction slot.
Claude Opus 5: Reproduction is trimmed from 2025's ten marks to a single 3–5 mark stimulus item, most likely a pregnancy or birth hormone graph, since fertilisation strategy was exhaustively examined in 2025 Q27.

Mod 7: Immune response and immunity 71% likely may rest

Chance of a big question (4+ marks) here: 39%

Question types predicted here short answer ×5 multiple choice ×3 stimulus based ×1

What each model expects

DeepSeek V4: Only a 3-mark question on memory cells or innate vs adaptive immunity will appear, not a large extended response.
Claude Fable 5: Immunity retreats to Section I plus at most a 3-4 mark written item after consecutive 7, and 11-mark treatments in 2024 and 2025.
Grok 4.6: Immunity appears as one or two MC items and at most a 2–3 mark innate outline; no 7–9 mark antibody-production chain after 2024–2025.
Claude Opus 5: Immunity is rested from a major Section II item after two heavy years, surviving as two or three multiple-choice questions plus a single short-answer item of no more than four marks.

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

Practise with the paper

Biology practice paper

100 marks · 35 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)

Exam Success Program

Book our ESPs now — online or in person

A paper on your desk is one thing — the clock running is another. Sit Biology practice exams under real conditions and get them marked with real feedback. Exam times run seven days a week, right up until the HSC begins.

Book your ESP →

Places are still open for 2026

Intu AI

One paper isn't enough? Generate more

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

Practise with Intu AI →

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.

The near-certainties

The pedigree — justify it, don't just name it

The epidemiology evaluation — validity, reliability, then a verdict

The transgenic-organism process

The big five (have these cold)

1. The pedigree — justify it, don't just name it

2. The epidemiology evaluation — validity, reliability, then a verdict

3. The codon-chart mutation question

4. Blood glucose negative feedback — the due data question

5. The transgenic-organism process

Also on the radar

Where to spend less time

The honest fine print

Where to spend your study time

Practise with the paper

Biology practice paper

Book our ESPs now — online or in person

One paper isn't enough? Generate more

Share & save