A free, open-source IQ estimation tool built as a single HTML file. No server, no tracking, no sign-up — just open it in a browser and go.
Take the test (GitHub Pages)
This test measures general cognitive ability (g-factor) across four domains using 35 timed questions. It's modeled after established psychometric assessments like Raven's Progressive Matrices, with difficulty-weighted scoring and z-score normalization to estimate an IQ range.
| Category | Questions | What it tests |
|---|---|---|
| Matrix Reasoning | 15 | Complete 3x3 visual pattern grids — shape, fill, size, rotation, and boolean logic rules |
| Pattern Sequences | 6 | Identify the next element in visual sequences with multiple changing properties |
| Number Sequences | 8 | Recognize numerical patterns (increasing differences, multiplicative, polynomial) |
| Logical Reasoning | 6 | Syllogisms, conditional logic, ordering, and constraint satisfaction |
Questions progress from difficulty 2 (moderate) to difficulty 5 (very hard). There are no "freebie" questions — even the easiest require identifying at least two simultaneous rules.
Scores are calculated using difficulty-weighted z-score normalization:
- Each correct answer earns points equal to its difficulty level (2–5)
- The weighted score is converted to a percentage of the maximum possible
- A z-score is computed against expected population parameters (mean = 0.44, SD = 0.18)
- The z-score maps to an IQ scale (mean 100, SD 15), clamped to 75–145
- A 95% confidence interval (±9 points) reflects the inherent uncertainty of self-administered tests
- Single file — the entire test is one
index.html, under 70KB. No dependencies, no build step - Fully offline — works without an internet connection once loaded
- Mobile responsive — optimized for phones with touch-friendly controls and horizontally scrollable question navigation
- SVG rendering — all visual questions are rendered as crisp, scalable vector graphics
- Anti-pattern safeguards — answer positions are randomized, distractors are psychometrically designed to each violate exactly one rule
- Timed — 25-minute limit with visual countdown and auto-submit
- Detailed results — bell curve visualization, per-category breakdowns, question-by-question review with correct answers shown
Visual question engine: A custom SVG shape renderer supports 11 shape types (circle, square, triangle, diamond, pentagon, hexagon, cross, star, arrow, dot, line) with properties for size, fill (none/gray/solid/striped), rotation, position, and stroke width. All shapes render into a normalized 90×90 viewBox that scales cleanly to any display size.
Hard questions use boolean logic on shapes:
- Subtraction — column 3 = shapes in column 1 minus shapes in column 2
- XOR — column 3 = shapes unique to exactly one of columns 1 and 2
- AND/Intersection — column 3 = shapes present in both columns 1 and 2
Distractor design: Each wrong answer violates exactly one rule from the pattern (wrong shape, wrong fill, wrong count, wrong size, etc.), making options plausible and preventing elimination by guessing.
It's a single HTML file. Deploy anywhere:
# GitHub Pages — just push and enable Pages in repo settings
# Or serve locally
python3 -m http.server 8000
# Open http://localhost:8000This is an informal educational tool. It is not a substitute for a professionally administered IQ test (e.g., WAIS-V, Stanford-Binet 5). Online tests lack controlled testing conditions, standardized normative samples, and clinical oversight. Results should be interpreted as a rough estimate, not a diagnosis.
MIT