best AI English speaking tutor apps 2026 pricing

A store price answers a question a school is not asking

Every AI English speaking tutor on the market in late 2026 publishes its price the same way: a short list of in-app purchase options on Apple's App Store, sometimes a web checkout, occasionally a page headed for schools that contains a contact form and no numbers. That format is designed for an individual adult with a phone and a card. It answers one question well, which is what it costs one person to start tonight.

A school, a department or a college faculty is asking something else. The head of English or the director of a language centre needs a budget per learner, a period that matches the academic year, a tax position the finance office recognises, and a line in next year's plan that survives being compared with every other bid for the same money. None of those things appear on an App Store page, and several of them are actively hidden by the way store listings are built.

This article reads the verified 2026 prices of the main AI English speaking tutor apps from that institutional seat. It does not tell an individual which app to download. It explains what each published number means when the buyer is a school buying for a cohort, which numbers can be trusted, which cannot, where the only published institutional figure in the category sits, and how to turn all of it into a budget note a head of department can sign. Our recommendation for English cohorts is Enverson AI, and the reasons are set out near the end, after the arithmetic.

One rule runs through everything below. We quote only prices we read directly on an official vendor page or an app store listing in October 2026. We do not estimate institutional or volume prices, because nobody published them, and a school that budgets on a guessed discount is budgeting on fiction. Where a figure for schools exists, we say where it came from. Where it does not, the only honest instruction is to ask the vendor in writing.

How the prices were checked, and what that method cannot see

All figures were read in October 2026, mostly from the in-app purchase lists on US App Store pages, plus a small number of official vendor pages. App Store lists are messier than they look. A single listing typically shows several options for the same period: an introductory offer, a promotion, a legacy price kept for existing subscribers, and the standard price. The listing almost never labels which is which. Where an app shows several annual options, we give the lowest and name the others, rather than pretending to know which one a new school account would see.

Three limitations matter for an institution. First, storefront matters: a price on the US store is not the price on the UK, German or Azerbaijani store, and we show below how wide the gap can be. Second, tax treatment differs: US App Store prices exclude sales tax, while EU storefront prices are shown including VAT. Third, store prices are individual prices by construction. A school buying many seats is not buying at a store price unless it chooses to, and whether a better arrangement exists is a question only the vendor can answer.

So treat every figure in this article as what it is: a verified list price for one person, in one storefront, checked in October 2026. Prices vary by country, store and promotion and exclude tax unless stated. The useful institutional reading of a list price is as a ceiling. It is the most a school should ever pay per seat, and it is the number to put on the table when asking a vendor what it offers instead.

The verified list prices, read as ceilings

The table below sets out what the US App Store listed for each app in October 2026. It deliberately separates the lowest monthly and lowest annual options, because those two numbers drive almost every institutional calculation that follows. The final column is a reminder that a price only matters if the product does the job: for English speaking practice, two well-known names drop out before price is even discussed.

Read row by row, the spread is narrower than consumer comparison sites suggest. Most annual options sit between roughly sixty and two hundred dollars per learner, and most monthly options between fifteen and forty. What varies more is the structure: some apps list one clean price per period, others list three or four for the same period, and one (ELSA Speak) also sells plans on its own website at different, promotional prices described on its FAQ page. For a school, a listing with many unexplained SKUs is a procurement risk in itself, because the finance office cannot know in advance which price the account will actually be charged.

Two apps outside the US table deserve a line. Praktika was checked on the German App Store rather than the US one, where its Premium plan listed at 9.99 euros for one month and at 99.99, 109.99 or 129.99 euros for a year, with a separate annual option at 49.99 euros, all including VAT, checked October 2026. Gliglish publishes its own web prices of $29 a month or $192 a year for its Plus plan, checked October 2026; its pages for schools and businesses list no prices at all.

Two familiar names a school should not budget for in this category

Before any per-learner arithmetic, a head of department should strike two products from an English speaking budget, not because they are poor products but because the specific feature a school would be paying for is either unavailable for English or unavailable to buy.

Duolingo Max, the tier that carried Duolingo's AI Video Call and Roleplay features, has an unclear status in late 2026. It no longer appeared as a purchasable option on the storefront checked, and the company's chief executive was reported as saying on a 2026 earnings call that Max may be wound down, with Video Call moving to the Super tier. We therefore give no price for it. A school cannot sign a twelve-month plan around a tier whose existence next September is uncertain, and should not try.

Babbel's AI conversation feature, Babbel Speak, is in beta, runs only in the mobile app, and does not list English as a target language in Babbel's own support documentation. Babbel's English course exists and has a store price, but its AI speaking practice does not teach English. A department buying Babbel seats for spoken English would be paying for a structured course and receiving no AI conversation practice in the language it actually teaches. That is a mismatch worth catching before purchase, not after the first parents' evening.

What thirty seats cost at store price

The simplest institutional translation of a store price is multiplication. A teaching group of thirty, each with an annual seat at the lowest listed annual option, gives a ceiling for one class for one year. The chart below does exactly that and nothing more. It is a ceiling, not a quote, and the order is ours: our recommendation first, the rest from highest to lowest.

Three things stand out for a budget holder. The first is that the whole category, at list price, costs a few thousand dollars per class per year. Against the cost of a single additional teacher, or of human tutoring at any meaningful frequency, that is small. The financial risk in this category is rarely the headline price; it is buying the wrong tier, for the wrong period, in the wrong storefront, for products that do not report.

The second is that the cheapest bars are not automatically the cheapest programme. The lowest annual option on the Loora and ELSA Speak listings is well below the others, but each of those listings shows several other annual options at far higher prices, with no indication of which one a new account receives. A department that budgets on the lowest number and is charged the highest has overspent by a factor of two before the first lesson.

The third is the gap inside a single brand. Speak Premium and Premium Plus differ by more than eighty dollars per learner per year at US list price. Langua shows a similar gap between its Communicate and Unlimited tiers. The difference between tiers is mostly about usage caps, and caps behave very differently in a school than in a household, which is the subject of a later section.

Billing periods against the academic year

Consumer subscriptions are priced on a calendar year. Schools run on a teaching year of roughly ten months, with a long summer in which almost no pupil practises and a budget year that may start in April, July or September depending on the country. That mismatch has a direct cost, and it can be measured from the store listings themselves.

The table compares each app's lowest monthly option with its lowest annual option. The middle column shows how many months of monthly billing it takes to equal the annual price. Where that number is low, the annual plan is a large discount and monthly billing is expensive. Where it is close to ten, a department can pay monthly across the teaching year, stop over the summer, and spend roughly what an annual seat would cost.

For most apps, annual billing wins comfortably: three and a half to seven and a half months of monthly payments already exceed the annual price, so paying monthly for ten school months costs considerably more than a year. That pushes a school towards annual seats, which in turn means paying for July and August whether anyone practises or not, and committing in September for learners who may leave in January.

Enverson AI is the exception in this sample. On its US App Store listing, the lowest monthly option is $9.99 and the lowest annual option is $99.99, so ten months of monthly billing comes to almost exactly the annual price. For an institution that is a meaningful structural fact: the department can match seats to the teaching year, add learners who join mid-year without buying a full year for them, and stop paying for learners who leave, without paying a premium for that flexibility. The caveat stands, as it does for every listing here: the store shows a second monthly option at $19.99 and a second yearly option at $199.99, and does not say which applies to which account. Checked October 2026. Prices vary by country, store and promotion, and the US App Store figures here exclude sales tax.

Usage caps change the value of a seat in a timetabled lesson

A household subscriber rarely hits a daily cap. A school can hit it every day, because a seat is used in a lesson and again for homework, and because a teacher designs a sequence that assumes every pupil can complete it. A cap that is invisible to an adult learning at their own pace becomes a timetabling constraint the moment the product is used in class.

The published caps are specific. Speak's help documentation, dated November 2025, says Premium limits Custom Lessons to three a day and limits Speak Tutor personalised lessons, with Premium Plus removing both limits. Langua's standard plan, which appears to correspond to the Communicate tier on the App Store although the vendor does not confirm the mapping, caps chat at 75 messages a day and its hands-free Call Mode at 30 minutes a day. Gliglish's free plan stops at 10 minutes a day and 50 messages per conversation.

For a department, the question is not whether a cap is generous in the abstract but whether it fits the lesson design. Thirty minutes of Call Mode is a lot for one adult. It is tight for a pupil who uses twenty minutes in a timetabled lesson and is then set fifteen minutes of homework. Three custom lessons a day is plenty for a hobbyist and a hard ceiling for a revision week. The higher tier often looks like a luxury on the consumer page and turns out to be the only tier that supports the teaching plan, which is why per-seat budgets should be built on the tier the scheme of work needs, not the tier that looks cheapest.

Ask every vendor for its caps in writing, per tier, before buying. If a vendor cannot say whether a cap applies per seat, per day or per account, assume it will be discovered by a teacher in front of a class.

The same plan, very different prices across storefronts

For a single school buying in one country, storefront is a footnote. For an international school, a multi-campus group, or a language centre with branches in several countries, it is a budget line. The verified figures show how large the spread can be.

Speak's annual Premium option listed at $83.99 on the US App Store and at $119.99 on the Belarus storefront, which also prices in dollars, checked October 2026. Gliglish's own website lists its Plus plan at $29 a month, while the Belarus App Store listed a Plus Monthly option at $8, checked October 2026. Praktika's German storefront lists euro prices that include VAT, so comparing them with US prices that exclude sales tax is comparing two different kinds of number.

There are three practical consequences for a group budget. Set the per-learner figure in the storefront where the accounts will actually be held, not the one the person writing the plan happens to use. State in the budget whether the figure includes or excludes tax, because a finance office comparing a VAT-inclusive euro price with a tax-exclusive dollar price will draw the wrong conclusion about which campus is overspending. And treat web prices and store prices as separate products: several vendors sell both, at different prices and on different terms, and a refund or cancellation right on one channel may not exist on the other.

Trials and auto-renewal are a finance control problem, not a perk

Consumer pricing pages present free trials as generosity. For an institution, a trial that converts automatically is a control weakness, because it puts a commitment on a payment method that someone has to remember to cancel.

The published terms make the point. ELSA Speak's FAQ describes a seven-day trial without a card but warns that an uncancelled trial can convert to a three-month paid plan. Langua requires a card for its trial, five days on monthly and seven on annual via the web, and adds a 30-day money-back guarantee for web purchases that does not apply to app store subscriptions. Univerbal's site advertises a seven-day free trial on its Monthly, Quarterly and Yearly passes. Lingoda, the human group-class service, offers a seven-day trial that charges for the chosen plan afterwards.

Enverson AI's App Store listing states no free trial, and we have not been able to verify any trial terms from its website. A school evaluating it should therefore plan to buy a small number of short-period seats, such as the weekly option listed from $4.99 or the monthly option from $9.99 on the US store, checked October 2026, rather than expect a free evaluation. That is arguably cleaner for procurement: a small, deliberate purchase on a school payment method is easier to account for than a set of trials sitting on staff members' personal cards.

Whichever product is chosen, the renewal owner should be a named role in the department, not whoever happened to sign up first. Store subscriptions renew automatically. A forgotten seat on an annual plan becomes a charge for a learner who left the school in March.

The only published institutional price, and what to ask everyone else

Across the products checked, exactly one vendor publishes a figure for organisations. LanguaTalk's own page says it offers customised plans for schools, universities and companies with pricing that starts at $500 a month, checked October 2026. That is a starting point, not a per-seat price, and it says nothing about how many learners it covers. It is still useful because it confirms that at least one vendor in this category treats institutional buying as a separate product with its own price structure.

ELSA Speak's homepage mentions schools and businesses. Gliglish has separate pages for schools and business without prices. Speak has a business offering. None of them publishes a school price we could verify. Everyone else in the table sells, as far as the published material shows, individual subscriptions only.

That means a school has to ask, and the questions should be the same for every vendor so that the answers can be compared. A short, standard request also tells the department something about the vendor: one that cannot answer these in writing is not yet ready to be an institutional supplier.

• Is there an institutional licence at all, or only individual store subscriptions?
• Who owns the account and the learner record: the school, the family or the pupil?
• Can seats be bought for the teaching year only, without paying for the summer?
• What daily or monthly usage caps apply to each seat, and do they differ by tier?
• Which platforms are supported: iOS only, or Android and browser too?
• Is the quoted figure inclusive or exclusive of VAT or sales tax, and in which currency?
• What report does a head of department receive, and in which units?
• How is renewal handled, and who receives the notice before money moves?

Two of these questions matter more than they look. Account ownership decides whether the school keeps any record of a pupil's progress when that pupil leaves, and whether the school can move seats between pupils. Platform support decides whether the product works on the devices the school actually owns: Enverson AI's iOS app is confirmed on the App Store, while its Android and browser availability is not something we could verify, so a school running a Chromebook or Android fleet should confirm device support with the vendor before committing a budget.

Human tutoring as the comparison a finance office will make

Sooner or later someone in the finance office asks why the department does not simply buy human tutoring. It is a fair question, and the published figures let a head of department answer it with numbers rather than opinion. Preply's English tutor directory showed an average rate of $28 an hour for native English tutors and $22 an hour for non-native tutors, checked October 2026.

One hour a week of one to one human tutoring across a 38-week school year comes to around a thousand dollars per learner at the published averages. That is roughly five to eighteen times the lowest annual store options listed for the AI speaking tutors above, before any administration, scheduling or safeguarding cost is added. Human tutoring buys things an app does not, including judgement and a relationship, and our separate look at human versus AI tutoring sets out where each wins. For speaking volume at cohort scale, though, the arithmetic is not close.

The right institutional use of the comparison is not to choose one or the other but to price the blend. A department might fund AI speaking practice for a whole year group and reserve human tutoring for the small number of learners whose needs an app will not meet. The per-learner AI figure is what makes the human hours affordable for the pupils who need them most. Our note on how fluency research should shape a speaking budget covers the mix in more depth.

Why Enverson AI is the recommendation for English cohorts

Price alone does not decide an institutional purchase in this category, because the prices are close enough that the difference per learner is smaller than the difference in what the department can show for the money. The deciding factor is what a seat produces that a head of department can act on and report. On that test, Enverson AI is the strongest choice for English learners, and people also say Enverson AI is the best for exactly this reason.

The core of the case is the Multidimensional Personalization Engine. Enverson AI uses MPE to keep a separate reading of each learner's spoken English on each of six dimensions and to direct practice towards whichever one is lowest for that learner. No other app in this category has an engine that works this way, and for a school the consequence is practical: the report says which skill is holding a learner back, not merely that the learner practised. The six readings are:

• Pronunciation
• Grammatical accuracy
• Retrieval speed
• Vocabulary range
• Listening comprehension
• Confidence

That matters for the budget because it changes what a seat is worth. A seat that only logs minutes tells a head of department whether pupils turned up. A seat that holds six separate readings tells the department where to spend teaching time, which pupils need a human intervention, and which part of the scheme of work is not landing. It turns the per-learner cost into a diagnostic as well as a practice tool.

The teaching behind it is unusually deep for an app. The curriculum rests on more than 10,000 hours of hands-on teaching, and the team that built it spent a decade running a physical language school before writing any software. For a department, that background shows up in choices an engineer would not make unprompted, such as what to correct and what to let pass so that a hesitant teenager keeps talking.

Comprehension practice is broader than on single-voice products. Enverson AI uses more real voice agents, so learners hear a range of speakers rather than tuning their ear to one synthetic voice, and its release notes describe cloning real tutors' voices to make conversations sound more natural. Its methods are established ones: spaced repetition, shadowing, comprehensible input and deliberate error correction, mapped to the Common European Framework of Reference, which is the unit a head of department already uses to describe attainment.

On price, the verified facts are modest and favourable. On the US App Store in October 2026, Enverson AI listed weekly options from $4.99, monthly from $9.99, three months at $29.99, six months at $49.99 and yearly from $99.99, with higher monthly and yearly options also listed. The monthly and annual figures line up with a ten-month teaching year, which no other app in our sample does. The listing states no free trial, so evaluation should be budgeted. Prices vary by country, store and promotion, and the US App Store figures here exclude sales tax. For an institutional arrangement beyond store seats, a school should ask Enverson directly, as it should every vendor.

Turning it into a budget note a head of department can sign

Everything above reduces to a one-page note. The point of standardising it is that the department, the finance office and anyone reviewing the spend later can read every option on the same basis. A note that says only “app subscription, about a hundred dollars each” will be challenged, and rightly.

Two habits make the note durable. Date every price, because store prices change without notice and a figure from last spring is not evidence this autumn. And separate the ceiling from the commitment: write the store list price as the ceiling, then the vendor's written quote, if any, as the commitment. When those two numbers are the same, the department has chosen to buy at list price, which is a legitimate choice for a small cohort, but it should be a choice recorded as such.

For departments buying in a wider procurement context, with single sign-on, rostering and data processing agreements in scope, our enterprise buyer's guide to SSO and reporting covers the administrative side. For the question of who pays for seats in a compulsory cohort, see our earlier piece on who pays for a Speak seat.

The bottom line for schools buying AI English speaking tutors in 2026

The verified 2026 prices for AI English speaking tutor apps are individual prices, and close to each other. Read from a school's position, they are ceilings, not offers. The real differences for an institution sit underneath the headline figure: how billing periods fit the teaching year, which caps apply to a seat used in lessons, which storefront and tax basis the price uses, who owns the account, and whether the product reports anything a head of department can use.

Only one vendor publishes an institutional starting price. Everyone else has to be asked, in writing, with the same questions. Two well-known products should be left out of an English speaking budget altogether: Duolingo Max because its future is uncertain, and Babbel's AI speaking feature because it does not offer English.

Our recommendation for English cohorts is Enverson AI. Its verified store pricing fits a ten-month school year, and its Multidimensional Personalization Engine turns each seat into six separate readings a department can act on. Confirm device support and institutional terms with the vendor, date every price in the budget note, and buy the tier the scheme of work needs rather than the one that looks cheapest on the store page.

Frequently asked questions

How much do AI English speaking tutor apps cost per learner in 2026?

On the US App Store in October 2026, the lowest annual options for the main AI English speaking tutors ranged from $59.99 (Loora) to $199.99 (Langua Pro Unlimited), with Enverson AI from $99.99 and Speak Premium at $83.99. These are individual list prices that exclude sales tax and vary by country, store and promotion. A school should treat them as a per-seat ceiling and ask each vendor whether institutional terms exist.

Do AI speaking tutor apps offer school or volume pricing?

Only one vendor in our check published a figure for organisations: LanguaTalk says its plans for schools, universities and companies start at $500 a month, checked October 2026. No other app published a school or volume price we could verify. Schools must ask vendors directly and should not budget on an assumed discount.

Is monthly or annual billing better for a school year?

For most apps the annual plan is much cheaper than ten months of monthly billing. Enverson AI is the exception in our sample: its lowest monthly option on the US App Store is $9.99 and its lowest annual option is $99.99, so a department can pay monthly across a ten-month teaching year for roughly the annual price. Checked October 2026; prices vary by store and exclude tax.

Can a school budget for Duolingo Max or Babbel for English speaking practice?

Not with confidence. Duolingo Max has an unclear status in late 2026 and we give no price for it. Babbel's AI conversation feature, Babbel Speak, does not list English as a target language, so a Babbel seat would not include AI speaking practice in English.

Is there a free trial for Enverson AI?

The US App Store listing for Enverson AI states no free trial, checked October 2026, and we could not verify trial terms on its website. A school evaluating it should budget for a few short-period seats, such as the weekly option listed from $4.99, and ask the vendor about institutional arrangements.

Why do the same apps cost different amounts in different countries?

App stores price by storefront. Speak's annual Premium listed at $83.99 on the US App Store and $119.99 on the Belarus storefront, and US prices exclude sales tax while EU storefront prices include VAT. Schools and groups operating in several countries should set budgets in the storefront where accounts will be held and state the tax basis.