AI language assessment in 7 minutes
Send a candidate a link. They speak and write in their own time, and a CEFR level comes back within minutes, with the ten scores behind it and the reasoning in plain English. 13 languages, nothing to schedule, from $0.60 a test.
No credit card. No sales call. No demo to sit through first.
Speaking & Writing · English
CEFR levelB2
Speaking
Writing
Held the thread of an argument across a long answer and recovered naturally after losing a word. Vocabulary was appropriate to a professional register throughout.
Hesitation before complex clauses slowed the pace. Article use was inconsistent in the written responses, though it never obscured the meaning.
- Speaking
- 7 min
- Speaking and writing
- 20 min
- Languages
- 13
- Scale
- A1 to C2
- From
- $0.60 a test
Read the transcript
On screen: The Evalingo logo, with "The AI language assessment platform" beneath it.
“Meet Evalingo, the AI language assessment platform.”
On screen: An employer's New assessment form. They choose English as the language, Speaking + writing as the test, and send the invite.
“Pick a language and a test, then send the invite.”
On screen: The candidate's invitation: their English assessment is ready, speaking and writing, about 20 minutes, start any time. They press Start assessment.
“Your candidate starts whenever it suits them.”
On screen: The same assessment running on a laptop and on a phone, recording a spoken answer.
“They can take it on any device, phone or laptop.”
On screen: A speaking question. A second voice reads it aloud while each word lights up as it is spoken, then a think-time countdown starts.
“Questions are read aloud, as well as shown on screen.”
Question, read aloud: “Tell us about a time you had to explain something complicated to a customer.”
On screen: A multiple-choice question is struck out and drops away. In its place, a candidate records a spoken answer and types a written reply to a client.
“No multiple choice. They answer open questions, out loud and in writing. It feels like a conversation, not an exam, and shows what they can really do.”
On screen: The thirteen languages, each written in its own script.
“In thirteen languages.”
On screen: A sample result: CEFR level B2 on the A1 to C2 scale, six scores between 68 and 76, and a note on what went well.
“Minutes later, you get a CEFR level, with the scores and the reasoning behind it.”
On screen: The Evalingo logo, "AI language assessment in 7 minutes", 13 languages, A1 to C2 CEFR, from $0.60 a test, and Start free at evalingo.ai.
“Evalingo. Start your free trial today.”
What is a CEFR proficiency test?
The Common European Framework of Reference for Languages (CEFR) is the international standard for describing language ability. Used by educators, employers, and language learners worldwide, CEFR provides a clear framework for assessing language proficiency levels from A1 (beginner) to C2 (proficient).
A traditional proficiency exam means booking a slot, travelling to a test centre and waiting weeks for a result. Evalingo returns the same CEFR level in about 20 minutes, from a link the candidate opens whenever it suits them, with nothing to schedule and nobody invigilating.
- A1Beginner
Fixed phrases and introductions. Cannot sustain an exchange.
Not viable for customer contact
- A2Elementary
Simple, predictable exchanges on familiar topics. Falls apart off-script.
Narrow scripted tasks
- B1Intermediate
Handles routine situations and explains what they mean, with visible effort.
Internal and back office roles
- B2Upper intermediate
Keeps up with a fast exchange, makes an argument, recovers when it goes sideways.
The usual bar for unscripted conversation with a customer.
Most customer-facing work
- C1Advanced
Fluent and precise. Handles nuance, tone and difficult conversations.
Escalations and client management
- C2Proficient
Effortless and exact, including implication and register.
Rarely worth requiring
Sample dashboard. Figures and names are illustrative.
Every result, in one place
Who you invited, who has finished, and what level each of them came back at. Sorted, filterable, and exportable over the API.
Language Assessment Dashboard
Overview of candidate proficiency tests
What the candidate sees
One question at a time, a visible clock, and nothing else on screen. Clear instructions, timed responses, and a smooth flow from start to finish.
- Mobile-friendly design works on any device
- Real-time audio recording with visual feedback
- Automatic progress saving for uninterrupted testing
- Available in 13 languages including English, Spanish, French, German, Swedish & more
Describe Your Ideal Work Environment
Speak for 80 seconds about your ideal work environment. Consider factors like team dynamics, workspace, and work-life balance.
Two skills, marked separately
Most roles need one of these more than the other, so you can run either on its own. Each is scored on five dimensions, weighted equally, and the level is what the ten add up to.
Speaking
5 to 7 minutes
Marked from recorded audio
- Pronunciation
- Fluency
- Grammar
- Vocabulary
- Coherence
Writing
15 to 16 minutes
Marked from real workplace tasks
- Task response
- Organisation
- Vocabulary
- Grammar
- Conventions
Both are scored by AI, with guardrails around it: responses are checked for attempts to talk the model into a better mark, results are flagged when the pattern of scores is implausible, and questions are drawn at random inside a fixed structure so two candidates never sit the same test. How the scoring works.
Find the version of this that fits you
Same assessment underneath. These walk through what it looks like for a particular job.
Language assessment software
What it is, what comes back, and how it compares to screening by interview.
Read more →BPO and contact centres
Screening agents at volume, checking the floor on a cycle, and proving a level to a client.
Read more →English proficiency testing
What each CEFR level sounds like, and which one your roles actually need.
Read more →13 languages, one scale
A B2 in Japanese means what a B2 in Spanish means. Each language has its own question library, and a candidate reads their instructions in the language they are being assessed in, not in English.
B2 is the usual bar for a role involving unscripted conversation with a customer.
Need one that is not here? Tell us which.
What happens, on both sides
- 01
Send a link
Create it in the dashboard or over the API. The candidate gets an email with their own link.
- 02
They take it
About 20 minutes, in the browser, whenever suits them. Each question is read aloud in the language being tested, so they listen and respond rather than only read. Nothing to book and nobody watching.
- 03
A level comes back
A1 to C2 within minutes, with ten dimension scores and written reasoning behind it.
- 04
You compare like for like
Every candidate sat the same structure against the same rubric, so the levels are comparable.
What the candidate meets
A timed test somebody is being scored on should be calm. Most of the work here went into removing reasons to panic.
- A microphone check first
- They confirm their audio works, and pick a different input if it does not, before anything is recorded.
- An introduction to each section
- What the section is for and how long it runs, so nothing arrives unannounced.
- Questions revealed as they read
- Text appears at reading pace rather than all at once, which stops the reflex to answer before finishing the question.
- Think time before recording
- 10 to 15 seconds to plan, every time, whether they want it or not.
- Nothing to press at the end
- Responses submit themselves when the time is up, so no one loses an answer to a missed button.
- Always visible progress
- Where they are and what is left, throughout.
From $0.60 a test
That is Professional billed annually, and it is the floor: nothing here is sold for less. The plans differ by volume, and volume is what moves the price per test. Everything else is the same on both: every language, both skills, the full API and webhooks.
Starter
For small teams getting started
100 assessments/month, $0.99 per test
Billed monthly, cancel any time
- 100 assessments included
- Speaking & Writing tests
- CEFR scoring (A1-C2)
- Full REST API and webhooks
- Dashboard access
- Email support
Professional
For growing companies
500 assessments/month, $0.75 per test
Billed monthly, cancel any time
- 500 assessments included
- Everything in Starter
- $0.75 per test, the best rate
- Top-up credit packs when you outrun the allowance
- Priority support
Enterprise
For large organizations
Custom volume pricing above 500 assessments/month
- Everything in Professional
- Volume pricing above the Professional allowance
- Invoicing rather than card payment
Both plans include every language, speaking and writing, the full REST API and webhooks. Ten free assessments to start, no card.
Full pricing detailQuestions people ask
- How long does it take?
- Speaking and writing together take about 20 minutes. Speaking on its own is 5 to 7 minutes and still returns a full CEFR level, which is all many roles need. Scoring comes back within minutes of the candidate finishing, and there is nothing to schedule.
- How does this compare to a human assessor?
- The honest claim is consistency rather than superiority. Every candidate answers the same structure with the same timings against the same rubric, which removes the interviewer to interviewer variation that makes human screening uneven. Every result shows its ten dimension scores and the reasoning behind them, so you can check a level you disagree with instead of taking it on trust.
- Which languages can you assess?
- 13: English, Spanish, French, German, Portuguese, Italian, Chinese (Mandarin), Japanese, Korean, Arabic, Russian, Dutch, Swedish. Each has its own question library, and all of them are scored against the same CEFR scale, so a B2 means the same thing whichever language it was taken in.
- Can it integrate with our ATS?
- Yes. Everything the dashboard does is available over a documented REST API: create assessments, poll status, pull back per answer scores. Webhooks fire the moment a result is ready, and a test mode lets your developers build the whole integration without spending credit or waiting on a real candidate. Pricing starts at $0.60 a test on annual billing, whether you drive it from the API or the dashboard.
Would you rather be walked through it?
You do not have to. Ten free assessments are a click away and nobody will call you. But if you would rather see it against your own roles and languages first, we will do that.
- A walkthrough mapped to your roles and languages
- Real speaking & writing assessments with AI scoring
- Plug into your ATS with our open API