Every credential. Proctored properly.
148 testing systems across language proficiency, higher-ed entry, professional licensure, aptitude and subject mastery — every one of them identity-verified, integrity-scored, and (where the test type allows) adaptively routed by IRT. No watered-down practice papers. The real instrument, run inside our ATLAS proctoring envelope.
The full instrument shelf.
Filter by category or search by name. Every entry is identity-verified at session start; adaptive tests route question difficulty based on running ability estimate.
Four layers of quiet enforcement.
Every session is wrapped in ATLAS. Findings are advisory — flags go to a human reviewer, never to automated penalty. Read the full spec at /products/atlas-proctoring.
Identity verification
Government-ID photo cross-checked against a live webcam capture using on-device face-landmark matching. Confidence threshold tunable per test sensitivity tier.
Read →Face & multi-monitor
Face-presence sampled at 4 Hz; off-frame events > 3 seconds are logged. Multi-screen and external display attachment detected via the Screen Capture API.
Read →Devtools detection
Console-open heuristics, focus loss, paste-event volume, and clipboard read attempts all stream to the session ledger. Each is a soft signal, never a hard fail.
Read →Integrity score
A composite 0-100 number combining face stability, focus loss, paste volume, and ambient audio. Below 75 routes the session to a human reviewer; above 95 is auto-released.
Read →The next question is tuned to you.
On adaptive tests, the engine maintains a running estimate of candidate ability theta after every response. It picks the next item whose difficulty b is closest to current theta — this is the most informative item under the 2-parameter logistic model.
As the response stream lengthens, theta tightens. Once the standard error of measurement drops below the test threshold (typically 0.30 in logit units), the session terminates and the score is locked.
Click Step to walk through one routing decision. Click Reset to start over. The orange branch is the path the engine actually took.
info(item) = a2 · P · (1 - P) → pick max info
Sessions running right now.
Anonymised candidate identifiers; integrity score is composite ATLAS output. Below-threshold sessions route to a human reviewer; the candidate is not notified mid-session.
| Candidate | Test | Minutes in | Integrity |
|---|---|---|---|
| CN-7F4A | Engineering aptitude · global | 42 min | 99 |
| CN-2E91 | English proficiency · academic | 118 min | 96 |
| CN-91BC | Graduate management aptitude | 64 min | 88 |
| CN-4F02 | Cognitive ability assessment | 18 min | 100 |
| CN-8D55 | Medical licensure · graduate | 152 min | 92 |
| CN-1A67 | Algorithms + data structures | 87 min | 94 |
| CN-6E20 | Japanese proficiency · N1-N5 | 95 min | 97 |
| CN-3C89 | Chartered accountancy · final | 175 min | 81 |
| CN-5B12 | Quantum mechanics mastery | 73 min | 99 |
| CN-9D71 | Banking probationary officer | 52 min | 95 |
| CN-7AC4 | Spatial reasoning · 3D rotation | 12 min | 100 |
| CN-2F38 | Civil services preliminary | 198 min | 90 |
Reserve a seat.
Slots open in 30-minute windows, 06:00 to 23:00 candidate-local. Special accommodations (extra time, reader, screen reader, sign-language interpreter, low-distraction room) are processed by a human within 48 h.
Four ways to sit a test.
Same instrument, same proctoring, same scoring. Only the fee path changes.
- Live + auto proctoring
- Score report within 5 days
- One reschedule free
- 1.5x or 2.0x extra time
- Screen reader / large print
- Sign-language interpreter
- Low-distraction room
- Income-based fee waiver
- Counter-signed by school principal
- Up to two tests per year
- Cohort-level dashboards
- API + SSO integration
- Custom score report templates
- Dedicated accommodation officer