中文（繁體）
中文（简体）
English
IB Maths AI HL Drilling Guide: Paper Order, Modelling Answers and a 12-Week Plan
TUTORZONE EXAM DRILL GUIDE · IB MATHEMATICS AI HL
Built on the published assessment framework: Paper 1 (technology allowed): Compulsory short-response on the whole syllabus; HL 110 marks, 2 (HL 30% / SL 40%); Paper 2 (technology allowed): Compulsory extended-response; HL 110 marks, 2 h, 30% (HL 30% / SL 40%); Paper 3 (HL only): Two compulsory extended problem-solving questions; 55 marks, 1 h (HL 20% (HL only)); Mathematical exploration (IA): 20% at both levels; 15 recommended hours (20% (SL and HL)). All three AI HL papers allow a graphing calculator, yet the marks sit in the model, the assumptions and the interpretation rather than the number. Method only: paper order, a twelve-week plan, four-line modelling answers, pacing, common losses and an error log.
How is IB Mathematics AI HL assessed, and what does each part carry?
Paper 1 (technology allowed): Compulsory short-response on the whole syllabus; HL 110 marks, 2 h, 30 (HL 30% / SL 40%); Paper 2 (technology allowed): Compulsory extended-response; HL 110 marks, 2 h, 30% (HL 30% / SL 40%); Paper 3 (HL only): Two compulsory extended problem-solving questions; 55 marks, 1 h, 20% (HL 20% (HL only)); Mathematical exploration (IA): 20% at both levels; 15 recommended hours (20% (SL and HL))
Assessment at a glance (2027)
| Component | Content | Weighting | Duration |
|---|---|---|---|
| Paper 1 (technology allowed) | Compulsory short-response on the whole syllabus; HL 110 marks, 2 h, 30% | HL 30% / SL 40% | 2 h |
| Paper 2 (technology allowed) | Compulsory extended-response; HL 110 marks, 2 h, 30% | HL 30% / SL 40% | 2 h |
| Paper 3 (HL only) | Two compulsory extended problem-solving questions; 55 marks, 1 h, 20% | HL 20% (HL only) | 1 h |
| Mathematical exploration (IA) | 20% at both levels; 15 recommended hours | 20% (SL and HL) | 15 recommended hours |
The official HL Paper 1 runs 120 minutes for 110 marks; this set’s 10 questions work out at about 12 minutes each. Marks in AI are rarely lost to arithmetic but to pressing the GDC before writing the model: a bare number with no equation, unit or contextual comment loses the method marks.
What each paper is actually testing
| Paper | Drill focus | How marks are lost |
|---|---|---|
| Paper 1 (technology allowed) | write the model before reaching for the GDC | Copies the calculator output without naming the model and its parameters, so method marks go |
| Paper 2 (technology allowed) | show the derivation and the interpretation | Applies a statistical test without checking sample and assumptions, inverting the conclusion |
| Paper 3 (HL only) | write the assumptions and route first | Splits Paper 3 time evenly between two long questions and finishes only the first half of each |
| Mathematical exploration (IA) | pick a question you can model with data | Chooses an exploration too broad to generate data, leaving nothing to reflect on |
What is each IB Mathematics AI HL paper really testing, and what should you drill?
Paper 1 (technology allowed): write the model before reaching for the GDC; Paper 2 (technology allowed): show the derivation and the interpretation; Paper 3 (HL only): write the assumptions and route first; Mathematical exploration (IA): pick a question you can model with data
Drill order: by topic, then by year, then the mark scheme
| Step | What you do | Why |
|---|---|---|
| 1. By topic | Work through each topic, starting with the paper that allows least help (no calculator / closed book) | Exposes what is genuinely not understood |
| 2. By year | Whole papers to time, one or two a week | Trains pace, stamina and question selection |
| 3. Mark scheme | Go through the marking scheme question by question and log the scoring steps you missed | Most lost marks are written wrong, not thought wrong |
Should drilling go by topic first or by year?
By topic first, then whole papers by year, then the mark scheme. The first third of the time goes topic by topic (8-10 years of the same topic back to back), the middle switches to timed whole papers, and the last fortnight only redoes the recurring errors from your log.
A 12-week drilling timetable
| Week | Phase | What to do |
|---|---|---|
| Weeks 1-3 | By topic: foundation | Topic by topic, starting with the easiest marks, building habits and recall |
| Weeks 4-6 | By topic: core | 8-10 years of the same topic back to back, errors classified by topic |
| Weeks 7-8 | By topic: advanced and electives | Elective and higher-order topics, now under time pressure |
| Weeks 9-10 | By year: timed papers | One full paper a week, timed, no reference materials |
| Week 11 | Mark scheme work | Question by question against the marking scheme; list the scoring steps you omitted |
| Week 12 | Plug holes and simulate | Only the recurring errors from your log, then one full exam simulation |
Subject-specific weighting for this timetable: Twelve weeks in three blocks: weeks 1-4 short answers with a mandatory four-line answer of model, assumptions, working and interpretation; weeks 5-7 statistics, naming population, sample and test conditions first; weeks 8-9 Paper 3 at thirty minutes per long question, banking marks first; weeks 10-11 timed papers; week 12 a full simulation and the error log. Two hours a week go to the exploration from week one.
Use it properly: this is not a checklist to finish. Every week must keep time for redoing errors, or from week 9 you will be repeating the same batch.
Pacing
| Paper | Time | Suggested split |
|---|---|---|
| Paper 1 (technology allowed) | 2 h | about 110 min with 10 min to check |
| Paper 2 (technology allowed) | 2 h | about 20 min per extended question |
| Paper 3 (HL only) | 1 h | about 26 min per question |
| Mathematical exploration (IA) | 15 recommended hours | topic, model, computation, interpretation |
How should the time be split inside each IB Mathematics AI HL paper?
Paper 1 (technology allowed): about 110 min with 10 min to check; Paper 2 (technology allowed): about 20 min per extended question; Paper 3 (HL only): about 26 min per question; Mathematical exploration (IA): topic, model, computation, interpretation
The drilling mistakes that cost the most marks
| Mistake | What it looks like | What actually happens | Instead |
|---|---|---|---|
| Copying the GDC display | Only 131000 is written | No model means no method marks | Write 250 000 × 0.85⁴ first, then the value |
| Wrong number of significant figures | Writing 130501.5625 | AI expects 3 s.f., so over-long answers lose marks | Write 131 000 and state 3 s.f. |
| Ignoring the data range in regression | Treating the x = 6 prediction as exact | A high r is not a licence to extrapolate | State the 1 to 5 hour range, then comment on the extrapolation |
The error log and the mark scheme
Four columns: where the mark went (model, assumption or interpretation), what the scheme wanted, next action (the fixed four-line template) and context card (one sentence on what the result means). Only faults seen three times earn the final fortnight.
How many years should you drill?
The current syllabus was first assessed in 2019; five or six sessions of official papers cover the main question types. Earlier mathematics papers belong to the previous syllabus and do not count as past papers for this course.
How many years of past papers is enough?
The current syllabus was first assessed in 2019; five or six sessions of official papers cover the main question types. Earlier mathematics papers belong to the previous syllabus and do not count as past papers for this course.
Frequently asked questions
Is this page a mock paper or a drilling guide?
This is the IB Mathematics AI HL drilling guide: assessment structure, drill order, a 12-week timetable, pacing, common mark losses and the error-log method. For questions with answers, use the IB Mathematics AI HL mock papers linked under Related.
By topic or by year first?
By topic. Eight to ten years of the same topic back to back is what exposes repeating errors; going straight to whole papers just repeats the same batch under time pressure.
How should the error log be written?
Do not copy out questions. Per error record four things: question reference, which step failed (concept, wording, calculation), what the mark scheme requires, and the correct action next time.
With a calculator allowed, do formulae still need memorising?
Yes, for different reasons. The GDC supplies values, but marks go to naming the model and saying why it applies, and that needs the conditions, not the key presses.
Official references
| Resource | What to use it for |
|---|---|
| IBO Mathematics AI subject brief (first assessment 2021) | Confirm the SL/HL weightings, durations and the technology policy |
| https://www.tutorzone.com.hk/en-ib-math-ai-past-paper-drill-guide/ | IB Maths AI drilling guide: modelling and interpretation |
| https://www.tutorzone.com.hk/en-ib-math-ai-sl-mock-paper/ | IB Maths AI SL mock (same series) |
Mock paper: IB Mathematics AI HL mock paper Practice: drill practice hub
Next step: find a tutor / register as a tutor
After drilling, the hard part is telling a conceptual error from a writing error. Tell us the year group, the school and your available times, and TutorZone will match you with a tutor who knows the HKEAA assessment requirements.
Demand for tutors in this subject is steady, and those who can explain the marking scheme and where marks are won are especially sought after. Register free; you pay only when a match succeeds.
