One million AI tokens a month. That is the ceiling on Carepatron's free plan, tucked three clicks into their pricing page, and in practice it covers between eight and twelve full clinical sessions before the scribe stops responding. Most Carepatron reviews in 2026 lead with "free AI scribe" and skip that translation.
Credit before critique. Jane App charges $15 a month for its AI Scribe. SimplePractice charges $35. Carepatron charges nothing and holds the line at a monthly token budget instead. Rare move in this category. Also the reason the free plan is a trap if you read "free AI" as "unlimited AI."
What most people get wrong about "free AI"
Priya, a solo acupuncturist in Kitchener, signed up for Carepatron's free plan during her slow January. She was not thinking about tokens. She was thinking about whether her evenings came back. Most clinicians I talk to read "free AI scribe" the way they read "free refills" — unlimited until the appointment ends.
Tokens do not work that way. A token is roughly three-quarters of a word, and a 45-minute visit burns a lot of them on both sides of the pipe: audio gets transcribed (input), then the transcript turns into a structured SOAP note (output). Each side draws from the same monthly budget. Follow-ups burn less. Dense intakes burn more.
The useful question is not "is it free." It is "how many sessions does my practice run per month, and where does the ninth session fall on the calendar?"
What is Carepatron's AI token limit on the free plan?
Carepatron's free plan caps AI usage at 1 million tokens per month, according to Carepatron's pricing page. One million tokens covers roughly 8–12 full clinical sessions, depending on visit length and note complexity. Once the cap is reached, the AI scribe pauses until the next billing cycle or a paid upgrade. Manual charting, scheduling, and invoicing continue to work unchanged — the platform does not lock the patient record.
The math that runs out by the second week
A million sounds like a lot. Then you divide.
A full 45-minute audio session, transcribed and summarized into a structured SOAP note, burns 80,000 to 120,000 tokens on the GPT-4-class models most vendors run under the hood. Oli's pipeline averages about 95,000 per typical visit. Carepatron has not published how they count, and we have not benchmarked theirs — read the numbers below as a ranged estimate, not a measurement.
Do the arithmetic. 1,000,000 ÷ 95,000 ≈ 10.5 sessions. Eight to twelve sessions per month, call it the ninth patient of the month plus or minus.
| Practice profile | Sessions / month | Estimated tokens at 95K / session | When the 1M cap hits |
|---|---|---|---|
| Part-time, ~5 sessions/wk | ~20 | 1.9M | Mid-month (around session 10) |
| Solo, 15 sessions/wk | ~65 | 6.2M | Week 1 (around session 10) |
| Solo, 25 sessions/wk | ~108 | 10.3M | First few days |
| 3-practitioner clinic | ~195 | 18.5M | Day 1 or 2 |
Priya tracked it. Week one of February: eight sessions, the scribe felt magical. Week two: the pane went quiet on her ninth patient of the month, mid-chart. She finished that note by hand at 8:40 pm, the way she used to, and remembered exactly why she had stopped doing that.
That is the window where most solo practitioners hit the cap. If you scribe every patient, the free plan is a week-one or week-two plan, not a full-month plan.
What happens when the meter hits zero
Credit where it's due. The tool does not break at the cap. It degrades cleanly.
The scribe pane says the feature is paused until next cycle and nudges you to Plus. Scheduling, manual charting, invoicing, and telehealth keep working. You are typing your own notes again, but you are not locked out of the patient record. Some "free tier ends" moments in other SaaS products put data behind a paywall. Carepatron does not.
Upgrading mid-cycle moves you to Plus immediately. Plus lists at $31 / month on their pricing page, with a limited-time 50% promo at $15.50 for the first six months before resetting to list. Annual billing trims another 20%. Advanced is $39 / month. Both paid tiers lift the token, appointment, and storage caps. On a three-year view, plan on the list price, not the intro rate.
Three scenarios, three bills
This is where "free" stops being the right word.
Scenario A — part-time clinician, 6 to 8 sessions a month. Free plan covers the AI side as advertised. The 100-appointment and 1 GB storage limits are the binding friction, not tokens. The catch: most part-time practices grow into a wall, not stay under one — by the time you are billing your tenth client a month, the scribe is already dark for the back half of the cycle.
Scenario B — solo clinician, 15 sessions a week. Roughly 65 sessions a month, scribe-on-every-visit: cap hits in the first 7 to 10 days. Two paths. Upgrade to Plus ($31 list, $15.50 promo) and keep AI on for the rest of the month, or treat the scribe as a week-one benefit and type the other three weeks by hand. Most solo clinicians who try path two end up on path one by month three.
Scenario C — three-practitioner clinic, roughly 180 sessions a month between them. Carepatron prices per user. Three seats on Plus is $93 / month at list, Advanced is $117, plus $39 / user / month if anyone e-prescribes. Flat-rate alternatives become meaningfully cheaper at this size. There is no free tier for a clinic. There is a free tier per clinician, and the bill scales with headcount.
What the free tier really gets you
| If you... | Then the free tier is… |
|---|---|
| See fewer than ~8 sessions a month | A free scribe today. Easy to outgrow once your caseload picks up. |
| See 10–20 sessions a week and scribe every visit | A week-one scribe. Budget for the Plus upgrade or move to a flat-rate platform. |
| Run a 2+ practitioner clinic | Not a free tier at the clinic level. Per-user pricing kicks in immediately. |
| Bill insurance electronically or e-prescribe | Add-on costs stack up ($0.19–$0.25 per claim, $39 / user for e-prescribe). |
The honest strengths, and the reasons to hesitate
A fair tradeoffs list in this category is rarely symmetric, and this one is not either.
What the marketing leads with. Cheapest credible option at the low end. AI included on every tier. Fourteen-day paid-feature trial with no card. Those are real, and they explain the sign-ups. They do not explain what happens in week three.
Where complaints stack up. Reliability shows up repeatedly on Capterra's public review board: crashes, delayed notifications, and a known absence of scheduling-conflict detection that has let double-bookings through. Reporting reads as thin. The UI feels older than its peers to some users. A practitioner I spoke with in Hamilton tracked her support tickets for a quarter and watched two double-bookings turn into refund requests. None of that makes the software broken. It does mean the failure modes are real and tend to bite as the practice scales.
Where Oli fits — a flat-rate Carepatron alternative built for working clinics
Oli Health is $19.95 per clinician per month, flat. The base price includes 2 million AI tokens — more than 40 hours of meeting transcriptions and the structured notes that come out of them — which is enough to cover a typical full-time month with headroom to spare. Past that ceiling, you pay only for the actual tokens you used. No surprise tier upgrades, no month-six promo reset, no per-feature add-ons. Scheduling-conflict detection is built in. Every AI output carries an audit trail the clinician can review and override, so the chart is yours, not the model's.
For a solo practitioner who currently hits Carepatron's wall in week two, that is the difference between a scribe that disappears mid-month and a scribe that stays on. For a three-practitioner clinic, it is roughly $60 a month at list, against Carepatron's $93 on Plus or $117 on Advanced — at clinic size, the math stops being close.
Switching EHRs is work, not a click, and pretending otherwise would be a lie. See how to choose an EHR for a solo practice for the questions worth asking before you migrate, and the broader pricing comparison for the category view. The 14-day Oli trial uses live patient data, not a sandbox, so the verdict on whether the workflow fits is yours by week two.
Is Carepatron really free for a working clinic?
The free plan is free in the literal sense — no card, no trial countdown — and includes AI scribe, scheduling, and charting. The limits make it unsuitable for most working clinics: 1 million AI tokens per month (about 8–12 full sessions), 100 total appointments, and 1 GB of storage. A full-time solo practitioner or anything multi-user hits a wall within days to weeks and needs the paid tier or a flat-rate alternative.
If you are on Carepatron's free tier right now, do one thing before Friday. Open the AI usage view and count the sessions you have already scribed this month. If you are past six, assume you will hit the cap inside the next two weeks. Decide now — upgrade, manual notes, or move to a flat-rate AI EHR — instead of at 9 pm on session twelve, mid-chart, with a patient's wait time ticking.
If a flat $19.95 with 2 million AI tokens included sounds better than a promo rate that resets at month six and a meter that drains in week two, start a 14-day Oli trial before your next Carepatron billing date. One login, one bill, and a scribe that does not go quiet on the ninth patient.
Quick questions
How many patients can I see on Carepatron's free plan?
Roughly 8 to 12 full clinical sessions per month before the AI scribe hits its 1-million-token cap. Short follow-ups and intake-only visits burn fewer tokens, so the count can run higher if your mix leans lighter. Separately, the free plan caps total appointments at 100 per month and storage at 1 GB, which tends to bind before tokens do for scheduling-heavy practices.
Is Carepatron reliable enough for a solo practice?
Reliability complaints appear regularly on public review sites like Capterra: crashes, delayed notifications, and a known gap in scheduling-conflict detection that has allowed double-bookings in some clinics. For a solo practice that depends on the calendar staying clean and the scribe staying on, the failure modes matter more than the price. Run any trial on live patient data, not demo data, before committing — that is the only test that tells you anything useful.
