How is a Custom Language Test different from the Adaptive test?

The Adaptive Language test detects a learner's CEFR level with one adaptive diagnostic that spans every band — you don't pick the level, it finds it. A Custom Language Test is the opposite: you author a fixed test to your own spec — your sections, your difficulty, your content, your branding. Use adaptive for placement; use custom when you want your own exam.

Custom Test Builder

Build Your Own Language Test Center

Author complete reading, listening, grammar, writing and speaking tests in 10 languages — under your brand. Pick a real-exam replica and auto-fill the whole form in one click, or paste your own passages and prompts in Fully Custom mode.

10 Languages 3 Scoring Outputs 5 Exam Replicas Your Content

English 🇬🇧 French 🇫🇷 German 🇩🇪 Spanish 🇪🇸 Chinese 🇨🇳 Arabic 🇸🇦 Hindi 🇮🇳 Japanese 🇯🇵 Portuguese 🇧🇷 Russian 🇷🇺

Launch Your Test Center See Adaptive Tests

Two language product lines

Adaptive detects a level. Custom builds a test.

Both run on one CEFR-aligned engine. Choose adaptive when you want placement; choose custom when you want your own exam.

Dimension	Adaptive Language Test	Custom Language Test
Core job	Detects CEFR level A1–C2	Authors a fixed test to your spec
Level selection	No pre-selection — genuine placement	You choose the target level & difficulty
Test shape	One adaptive engine, single section	Any sections you pick, fixed form
Content source	AI-generated across every band	AI-generated or your own passages/prompts
Result format	CEFR placement	Points · CEFR · Native band
Exam replicas	—	IELTS / TOEFL / PTE / OET / CELPIP auto-fill
Best for	Placement, streaming, entry testing	Branded exams, exam prep, corporate/immigration

Looking for automatic CEFR placement instead?

The Adaptive Language Proficiency test finds a learner's level A1–C2 in minutes with a single adaptive diagnostic — no level to guess up front.

Explore Adaptive Language Testing

One builder. Two ways to work.

Start from a real exam and tweak, or build every section from scratch.

Standard-Based Replicas

Pick a Base Test Style and the entire form auto-fills to real-exam shape on page load — sections, durations, passage counts and question types all pre-set. Override anything you like.

IELTS — 3 reading × 13 Q · 4 listening · 2 writing · 3 speaking
TOEFL — 2 reading · 5 listening · 2 writing · 4 speaking
PTE Academic — 3 reading · 5 listening · 2 writing · 5 speaking
OET — referral-letter writing · role-play speaking
CELPIP — 4 reading · 6 listening · Canadian everyday topics

Fully Custom

Turn auto-fill off and build every section your way — any language, any format, and your own content. Paste it in and the AI does the rest.

Reading — paste a passage → AI writes MCQ, True/False, fill-blank
Listening — paste a transcript → AI writes questions and generates TTS audio
Writing — supply a prompt → AI grades against a calibrated rubric
Speaking — supply a prompt → student records, AI evaluates
Grammar & Vocabulary — dedicated MCQ section (unique to custom)

One test. Three ways to score it.

The same test can report a raw score, a CEFR level, or a real-exam band — you choose per template.

Numeric Points

The universal default. Students see 72 / 100 or 72% with a letter grade. Perfect for corporate and internal assessments where you just need a number.

CEFR Level (A1–C2)

A placement badge — B2 · Upper Intermediate — with a per-skill CEFR breakdown, an A1–C2 axis and a "what your level means" table. Ideal for placement and streaming.

Native Scale

The chosen exam's own band — Band 6.5, 70/90, Grade B, 87/120 or CLB 8 — with per-skill native scores and a scale explainer. Best for exam prep.

Honest by design

CEFR and Native results carry a difficulty-aware calibration note on the page and PDF — a calibrated benchmark, clearly framed as an internal estimate, never passed off as an official proctored result. That keeps your institution's claims defensible.

Build → Generate → Deliver

From a blank form to an assignable, AI-graded test — in one sitting.

Configure

Pick language, difficulty and builder mode. Choose sections and the scoring output — or auto-fill a replica.

Generate

AI builds every section in the background — questions, audio and rubrics. Keep working; the page tracks progress.

Assign

Publish to /ai-exams/, assign to students, or run a placement flow where each student sees only their test.

Score & Report

AI grades speaking and writing; results render in your chosen output mode as an on-screen report and branded PDF.

Who builds custom tests

Language schools

Placement and progress tests at exactly the CEFR level you teach — reading + grammar only if that's all you need.

Corporate HR

Screen candidates on job-relevant language with a simple points score and your own scenario content.

Immigration centers

Run native-scale mocks — IELTS Band, CELPIP CLB — to benchmark readiness before the real exam.

Universities

Branded entry assessments in multiple languages, delivered and graded on your own white-labeled platform.

10

Languages

5

Exam replicas

3

Scoring outputs

5

Section types

Frequently asked questions

The Adaptive test detects a learner's CEFR level with one adaptive diagnostic across every band — you don't pick the level, it finds it. A Custom test is the opposite: you author a fixed test to your own spec — your sections, difficulty, content and branding. Use adaptive for placement; use custom when you want your own exam.

Ten: English, French, German, Spanish, Chinese, Arabic, Hindi, Japanese, Portuguese and Russian. Arabic is rendered fully right-to-left across every section with native TTS audio.

Yes. In Fully Custom mode every skill section accepts your own content — paste a reading passage, a listening transcript, or a writing/speaking prompt, and the AI generates questions (and TTS audio for listening) from exactly that content.

Three: raw Points (72/100 or 72%), CEFR Level (an A1–C2 placement badge with per-skill breakdown), or Native Scale (IELTS Band 6.5, PTE 70/90, OET Grade B, TOEFL 87/120, or CLB 8). CEFR and Native modes include an honest, difficulty-aware calibration disclosure.

Generation runs in the background. A typical test is ready in about 5–15 minutes; a full five-section multilingual test can take 40–55 minutes. You keep working while it generates and the page polls for completion.

Launch your own language test center

Branded, AI-graded tests in 10 languages — real-exam replicas or fully your own.

Talk to Us View Pricing