PILLAR 01 Foundation · rubric-aligned · 11 subjects live · 60+ planned

Grounded in the actual IB rubric.

Generic AI guesses at grades. iBMarker is calibrated criterion-by-criterion against official IB assessment rubrics - so every score, every comment, and every level descriptor speaks the language your moderators expect.

Get Started · Free Back to pillars
Why rubrics

Why rubric alignment is non-negotiable.

The difference between a useful AI marker and a misleading one isn't model size - it's whether the grader speaks IB.

01 - Criterion-level marking

One score per rubric criterion, not one blob.

Every IB rubric breaks an answer into discrete criteria - Knowledge, Application, Analysis, Evaluation, Synthesis - and assigns level descriptors with explicit mark bands. iBMarker scores each criterion independently, just as a moderator would, then composes the final grade from those slices. Students see exactly which criterion cost them marks, with the level descriptor they fell short of and a quote from their own answer as evidence.

What you get

Per-criterion
Independent scores, not a single number.
Level descriptors
Official IB language quoted back to the student.
02 - Command-term awareness

"Evaluate" and "describe" are not the same task.

IB command terms - state, define, describe, explain, analyse, evaluate, justify, discuss - dictate the depth of response expected and the rubric criteria that apply. iBMarker parses the command term from every question and grades accordingly: an "evaluate" answer is held to the higher-order rubric bands, while a "state" answer is checked for precision rather than depth. Students who write an essay for a 2-mark "state" question get told why - they don't lose marks silently.

What you get

20+ command terms
Each mapped to the rubric depth it requires.
Calibrated depth
No more "wrote too much for too few marks".
03 - Calibrated to teacher consensus

0.94 correlation with human teachers.

Each subject's marker is trained on thousands of teacher-marked exemplars sourced from IB schools across the network, then validated against held-out moderator scores. Current performance: 0.94 score correlation with human teachers, 68% exact-mark agreement, and 91% within plus-or-minus 0.5 marks. The marker is re-calibrated every IB cycle as new subject reports and updated mark schemes are released - so accuracy doesn't drift.

What you get

0.94
Correlation with human teacher marks.
Re-calibrated
Every IB cycle, every syllabus change.
04 - Syllabus-grounded, not rubric-scaffolded

Questions and answers anchored in the syllabus.

Most "IB-aligned" AI tools just paste the rubric into a prompt and hope. iBMarker is grounded in the actual IB subject guides - syllabus content, assessment objectives, and topic-level learning outcomes - so a generated question on, say, monetary policy quotes the right syllabus section and the marker checks against the specific outcomes that section covers. The result: questions and feedback that wouldn't look out of place in an IB-issued past paper.

What you get

Syllabus-anchored
Questions and marks tied to subject guides.
Defensible
Every score traceable to a learning outcome.
Live coverage

11 subjects live, 60+ on the roadmap.

Full criterion-level AI marking is live for the subjects below. Assignment distribution, gradebook, and analytics work for any IB subject - even those still in the marker queue.

Sciences

  • Biology HL & SL
  • Chemistry HL & SL
  • Physics HL & SL
  • Computer Science HL & SL

Individuals & Societies

  • History HL & SL
  • Economics HL & SL
  • Business Mgmt HL & SL
  • Psychology HL & SL
  • Geography HL & SL

Mathematics

  • Analysis & Approaches HL & SL
  • Applications & Interp. HL & SL
iBMarker
No subscription. No commitment.
Full Access.
Free

Try rubric-aligned AI marking on your class.

Onboard your IB department in minutes. Full marker, gradebook, gaming layer, and analytics - free for schools, teachers, and students.

View Detail Get Started.