Three AI tools price a school year in three different units

The classroom note

A finance officer asked what an AI tool costs "per student per year" usually cannot get a straight answer, because three widely used tools price themselves in three different units, on three different terms. Reading their published pricing and pricing-adjacent pages as retrieved on 16 September 2026 shows why the question is harder to answer than it sounds.

What the evidence says

Khan Academy's own blog post on its Khanmigo assistant states that its "content library and learning platform will always remain 100% free," while individual US adult access to the AI tutor itself carries a "$4/month or $44/year payment." MagicSchool's pricing page prices per teacher, not per student: a Free tier at "$0 Forever," a Plus tier at "$8.33 USD/user/mo (billed annually)" or "$12.99" billed monthly, and an Enterprise tier with no published figure, only "Get pricing." The same page compares itself to rivals it says cost "$25 a month" (Microsoft Copilot) and "$20 per month" (ChatGPT Plus) - a vendor's own comparison, not an independent one. OpenAI publishes no per-seat figure for its institutional ChatGPT Edu product at all, offering only "Contact Sales," while a separate consumer-facing plan, ChatGPT for Teachers, states it "is free through June 2028 for verified U.S." K-12 educators.

The implementation question

None of these figures answers "per student per year" directly. Khan Academy's $44 is an individual subscription, not a school licence; MagicSchool's $8.33-$12.99 is charged per teacher seat, so a school's per-student cost depends entirely on how many staff it licenses relative to its enrolment; and neither OpenAI figure is a standing institutional price at all - one is unpublished, the other a time-limited offer that expires in 2028. A district building a real per-student figure has to convert each of these into the same unit itself, and separately ask what happens to the price, and the tool, once a free or discounted period ends.

What holds and what fails

What holds: publishing any concrete number, as Khan Academy and MagicSchool both do for their smaller tiers, is more checkable than a "contact us" page, and it is fair to compare tools on that basis where the numbers exist. What fails: comparing a per-teacher price to a per-student price as though they were the same measurement, or treating a 2028 expiry date as a stable long-term cost. This is an editorial comparison of published pages, not a procurement quote, and enterprise or district pricing for any of these tools may differ from what is shown here.

• Is this price per student, per teacher, or per institution, and how does that convert to our actual enrolment?
• Does the free or discounted tier have an end date, and what does the tool cost once it passes?
• Where a page says "contact sales," have we actually obtained and compared a written institutional quote?

A published number is worth more than a sales call precisely because it can be checked. Where none exists, that absence is itself worth recording before a contract is signed.