How Much Does Synthesia Cost? Free, Starter and Pro for a Training Library

Synthesia costs $0 on the free Basic plan, then $19 a month for Starter ($14 a month billed yearly) and $89 a month for Pro ($59 billed yearly), on the pricing page as I read it on 5 November 2026. For a small training library, Starter covers a handful of short lessons; Pro suits steady production.

Check the figures against your own screen before you budget. Pricing pages like this one change without notice, and the page currently advertises first-month offers ($9 for Starter, $49 for Pro) that make the first invoice look cheaper than every later one. Everything below uses the ongoing price, not the offer.

Follow me on Instagram@sagnikteaches

The four plans as the pricing page lists them

The page shows monthly billing by default and a yearly toggle. Usage is measured in credits, one shared pool for video, dubbing, surveys and the AI video agent called Syren, rather than in separate minute allowances.

Connect on LinkedInSagnik Bhattacharya

Two details trip people up. First, every paid plan still includes one editor; the "guests" are reviewers who can comment, not colleagues who can build. Second, the free allowance (10 minutes plus 500 credits) sits on top of paid credits, so a Starter account effectively has a little more than the headline number.

Subscribe on YouTube@sagnikteaches

Turning credits into minutes you can plan with

The page translates credits into approximate video length: about 12 minutes a month on Starter and about 60 on Pro, in both cases on top of the free 10 minutes. Divide it out and Studio video works out at roughly 100 credits a minute. Dubbing is a little cheaper per minute (up to 16 minutes on Starter, 78 on Pro), and survey responses and Syren videos draw on the same pool.

Syren is the exception to the 100-a-minute rule of thumb. The page says a one-minute Syren video typically uses 1,000 to 2,000 credits, so a single minute of AI-agent video can eat most of a Starter month. Pick the Studio editor for lesson libraries and treat Syren as an extra for the odd experiment.

My arithmetic here is an estimate from the page's own approximations, not a published rate card. The sensible test is to render one real 60-second lesson on the free plan and watch how far the credit counter drops.

A 24-lesson onboarding library, costed month by month

Take an illustrative bakery with 14 staff that wants an induction library: 24 lessons of about three minutes each (food safety, allergen labelling, till procedures, opening and closing checks), so 72 minutes of finished video. Allow a third as many minutes again for re-renders after edits, and the generation budget is about 96 minutes.

The gap is smaller than the sticker prices suggest. Starter is cheaper per month but stretches the project across five months, and every month you wait is a month new hires are watching nothing. Pro finishes the library inside two billing cycles for roughly twice the cash, and you can drop back to Starter or Basic for maintenance afterwards. If the library is the deadline, Pro for two months is the cheaper route in real terms; if staff are happy to receive lessons in instalments, Starter wins.

Whichever you pick, check whether leftover credits roll over. The page does not say so, which is why the table assumes they do not.

What the first month looks like on Starter

With about 22 minutes available, the bakery should not start at lesson one. Start with the highest-risk lessons, the ones where a wrong action costs money or health, so the first month produces the videos that matter most.

1. Allergen labelling and cross-contact (4 minutes).
2. Opening and closing checks (3 minutes).
3. Till and refunds procedure (3 minutes).
4. Oven safety and burns first aid (4 minutes).
5. Delivery acceptance and temperature log (3 minutes).
6. Complaint handling script (3 minutes).

That is 20 minutes, leaving two minutes of headroom for one redo. Lessons seven to twenty-four wait for months two to five, in the order that new starters need them.

A script edit that saves credits

Credits are spent on rendering, not on typing, so most of the saving comes from getting the script right before you generate. Here is an illustrative before and after for one lesson.

Before, copied from a handbook paragraph:

All employees are required to ensure that prior to the commencement of
each shift they have reviewed the allergen matrix, which is located in
the folder at the front counter, and verified that all products on
display are correctly labelled in accordance with the matrix.

After, rewritten to be spoken in about 20 seconds:

Before every shift, open the allergen folder at the front counter.
Check each product on display against the matrix. If a label is missing
or wrong, take the product off the shelf and tell the duty manager.

The second version is shorter, uses commands instead of passive phrases and ends on an action. It also avoids a classic cost trap: a long sentence with three clauses tends to get re-recorded because the pacing sounds wrong, and each re-render costs about the same as the first. Read every script aloud once and cut anything you stumble over.

The re-render habit that drains a month

The mistake that shows up in most trial accounts is generating the whole video to check a single scene. A bakery owner changes the fridge temperature from 5 degrees to 4 in the cold-storage lesson, renders the full three minutes, notices a misread number, fixes it and renders again. That is six minutes of credits for a one-word change.

Two habits stop it. Preview scene by scene in the editor before the final render, and keep a "numbers and names" list on one page (temperatures, product names, shop hours) that you check against the script before you press generate. A pre-render checklist takes two minutes and saves a month's worth of credits across a library.

Keeping the library current without rebuying it

Training libraries are never finished. Suppose four of the 24 bakery lessons change after a new allergen supplier and updated till software: about 12 minutes of video, or roughly 1,200 credits. That is one Starter month, not a new library. The ongoing cost of a library you already built is therefore small if you update in batches, which argues for building it on Starter or yearly billing and not staying on Pro.

For a library that changes weekly, such as one for a fast-moving sales team, the arithmetic reverses, because the credits are used every month. If you regularly regenerate more than about 20 minutes a month, Pro's 60 minutes becomes the plan that matches your usage.

Yearly billing against monthly: the sum

Yearly billing saves $60 on Starter (12 x $19 = $228 against $168) and $360 on Pro (12 x $89 = $1,068 against $708). The saving is real, but it only pays if you would keep the plan for the whole year. A simple rule works.

• Monthly billing if the library will be built and then left alone, or you are still unsure Synthesia is the right tool.
• Yearly billing if you know you will update lessons every month and you have already rendered a test video you were happy with.

A sensible middle path is two monthly cycles to build, then a switch to yearly once the plan has proved itself. Check that switching mid-term is allowed on your account before relying on it.

Who should stay on the free Basic plan

Basic is worth more than it looks as a drafting room. You can test scripts, compare avatars and voices, judge whether staff will accept a presenter that is not a person, and see exactly how many credits a real lesson consumes. It stops being useful the moment you want to hand a file to a new hire, because videos are shared by link only and carry the watermark.

My advice is to spend one afternoon on Basic with a single lesson before paying anything. If you decide the format does not fit your team, the test cost nothing. The longer free-plan walkthrough is in what the Basic plan allows.

What Pro adds beyond more credits

Pro raises personal avatars from three (on Starter) to five and adds a larger stock library (180+ against 125+), branded video pages, interactive videos, five guests for review and priority chat or email support. Personal avatars are the part to treat carefully on either plan: a personal avatar is made from a real person, so you need written consent from that person and a clear plan for what happens to the avatar when they leave. The consent steps for personal avatars cover that in detail.

Interactive videos matter if a lesson ends with a question and you want branching, but most onboarding libraries do not need them. Branded video pages are useful if you share lessons outside your learning platform or intranet.

Other costs that sit outside the Synthesia invoice

• Script time. A three-minute lesson needs about 450 spoken words and an hour of drafting and checking. Twenty-four lessons is therefore around 24 hours before anyone presses generate.
• Review time. Someone who knows the procedure should watch every lesson once. Budget 10 minutes a lesson.
• Hosting. The share link is fine for a small team. Video embeds into a website or learning platform are listed on every plan, but SCORM export, which most learning platforms need for completion tracking, appears only on the Enterprise card, so confirm what your platform accepts before you plan around it.
• Translation. Dubbing draws on the same credits. One-click translation into 140+ languages is listed on the Enterprise card, so on Starter or Pro you would budget dubbing minutes yourself.

When owners compare costs honestly, the hours spent on scripts and reviews usually exceed the plan fee. That is a reason to write each lesson once, properly, and to compare Synthesia with cheaper formats such as Canva for lessons that change often.

The same sums for an HVAC contractor with a technician course

A different shape of library: an illustrative HVAC contractor with eight technicians wants twelve ten-minute refresher modules on refrigerant handling, van stock checks and customer-site etiquette. That is 120 minutes of finished video, far longer per lesson than the bakery's, so the plan question changes.

At roughly 100 credits a minute, 120 minutes needs about 12,000 credits. Starter's 1,250 a month would take almost a year; Pro's 6,000 takes two months. The contractor's real choice is therefore not Starter against Pro but whether ten-minute avatar lectures are the right format at all. Long lessons are where avatar video tires viewers and burns credits fastest, so the better plan is to cut each module into three-minute clips, one per task. Thirty-six short clips cost the same credits, but each one can be updated alone when a van checklist changes.

The lesson for any business: count finished minutes, not lessons, and split anything past four minutes before you price it.

A budget sheet to fill in before you subscribe

Copy this into a spreadsheet and replace the example figures with your own. The example column is the bakery from earlier.

Lessons planned ................................ 24
Average finished minutes per lesson ............ 3
Finished minutes (lessons x minutes) ........... 72
Re-render allowance (one third of the above) ... 24
Generation minutes needed ...................... 96
Credits needed (minutes x 100) ................. 9,600
Plan credits per month (Starter 1,250) ......... 1,250
Free allowance per month (about 10 min + 500) .. about 1,500 credits
Months to finish ............................... 9,600 / 2,750 = 3.5, round to 4 or 5
Cost at the monthly price ...................... months x $19

The "free allowance" line converts the 10 free minutes to roughly 1,000 credits and adds the 500 free credits. If that conversion feels generous, drop the 1,000 and keep the 500: the sheet then says six months on Starter (9,600 / 1,750 = 5.5), which makes Pro look better. Run the sheet both ways, then decide with the worst case in view.

Three owners, three plans

A quick way to pressure-test the choice is to run three typical situations through it.

• A solar installer with a one-off safety induction. Eight clips of two minutes, 16 minutes in total, rarely updated. Starter on monthly billing for one or two months, then cancel. Cost: $19 to $38, or $9 plus $19 with the first-month offer.
• A bicycle repair shop adding a "how to" series for customers. Twelve one-minute clips, new ones monthly. Starter on yearly billing, because the stream of short clips fits inside 22 minutes a month. Cost: $168 a year, about $14 a month.
• A bakery chain adding a new site every quarter. Each opening needs the same 24 lessons customised for local menus and a new manager. Pro, because the volume is high and recurring, ideally with yearly billing after the first two months prove it. Cost: $708 a year if billed yearly.

Notice that the decision follows the shape of the work, not the size of the company.

Sample trial test with a pass mark

Before paying, run the free plan through a short test and write down the results. An illustrative test sheet for one lesson looks like this.

Lesson: Opening checks (script 160 words, target 60 seconds)
Avatar chosen: stock presenter, neutral office background
Credits before render: 500
Credits after render:  ___   (record the difference)
Render time: ___ minutes
Pronunciation errors: ___ (list product names that were misread)
Would a new starter follow it without help? yes / no
Would the owner put their name to it? yes / no

Pass mark: the credit drop is close to your 100-a-minute estimate, there are no more than two pronunciation fixes, and both answers are yes. If the credit drop is much higher than expected, multiply your budget by the same ratio before you pick a plan. A test like this takes under an hour and prevents the common outcome of buying Starter, discovering at lesson six that every lesson needs three renders, and moving to Pro anyway.

What to do when the pronunciation is wrong

Product names, surnames and trade terms are the usual casualties. A hypothetical misread is a brand of oven or a cleaning chemical that the voice stresses oddly. The fix is cheap if made in the script: respell the word phonetically in the narration text and keep the correct spelling on screen. Do that on the free plan during testing and keep a short glossary, so the same fix applies across all 24 lessons instead of being discovered one render at a time.

Checking your own bill after the first month

When the first invoice arrives, confirm four things. The plan name matches what you chose; the amount is the ongoing price (the first-month offer is gone from the second invoice); the credit counter shows what you used; and the billing screen says whether unused credits carry forward. If anything differs from this tutorial, trust your screen and Synthesia's help centre, not me, since prices and allowances are a moving target.

If a lesson library is part of a larger plan to put AI into onboarding, the AI guide places video next to the other jobs worth automating, and turning a handbook into short videos shows the production side once the plan is chosen.

Synthesia cost questions from owners planning a training library

Can I build a whole training library on the free Basic plan?

Only a small one, and not to publish. Basic gives 10 minutes of video a month plus 500 credits, with a Synthesia watermark and no downloads. It is a test bench for scripts, avatars and voices. Real lessons for staff need a paid plan to remove the logo and download files.

Do unused credits roll over to next month?

The pricing page does not say, so do not budget on it. Treat each month's credits as use-it-or-lose-it until Synthesia's help centre or your billing screen tells you otherwise, and schedule heavy production into the months you have paid for.

Is Pro worth it just for the five personal avatars?

Rarely. Starter already lists three personal avatars and Pro raises that to five, so the avatar count only matters if more than three people need a digital twin. Each personal avatar needs consent from the person filmed, and stock avatars cover most staff training. Pro is easier to justify on volume: about five times Starter's monthly credits.

What does a second editor seat cost?

Starter and Pro each list one editor plus guests (3 on Starter, 5 on Pro) who can comment. Extra editors are not priced on the page, so a team that needs two people building videos should ask Synthesia for a quote before committing.

Further reads

• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Synthesia pricing page (synthesia.io/pricing), fetched 5 November 2026; the plan figures are USD list prices.