Synthesia Pricing in 2026: What a Marketing Team Actually Pays

What a Synthesia AI video generator actually bills you for

This is the table that should reorganise how you read the plan ladder, because Synthesia does not sell video by the minute. It sells credits, and the credit's value swings by a factor of nine depending on which feature spends it. The list lives in the pricing page's FAQ and the help centre, never on the plan cards.

Source: synthesia.io/pricing FAQ and help.synthesia.io article 11972627, both read 2026-10-04.

Now convert one pool into four different products, because nobody else does. 1,250 credits is 12.5 minutes of typical video, or 31 minutes of dubbing without lip sync, or 15.6 minutes of lip-synced dubbing, or 3.6 survey responses. One allowance, four completely different months of output, and the "10 minutes included" headline that every explainer prints is one of those four cases in its most generous form.

And the vendor does not agree with itself on the top rate. The pricing FAQ and the help centre both say about 100 credits per finished minute. The vendor's own product announcement says 120. Its own help-centre edit example implies 120 (320 credits for a two minute forty second video). Its own ad copy on the adjacent query says "$0.40 per minute". That is a 20% spread between the vendor's knowledge base and the vendor's announcement feed, on the single number a buyer needs to plan a month. At 100 credits a minute, Starter's allowance covers 12.5 minutes of video. At 120, it covers 10.4. Use 100 as your planning basis, not because it is right but because it is the figure on the page a buyer will actually read, and it is the more conservative of the two for capacity.

One rule here earns the vendor real credit, and I want to be fair about it. Consumption is based on "the avatar footage generated, not the length of the finished video", and edits bill only new footage: "edit three seconds of an existing video and regenerate, and you're charged for three seconds." Removing words costs nothing. That is the opposite of the retake tax that the equivalent avatar-video pricing record carries, and it means a screen-recording or slide-led video stretches the same allowance considerably further than an avatar-led one. Two teams with identical output length can have materially different bills, and the difference is how much avatar time is on screen.

Synthesia pricing per finished minute, at every tier

Here is the conversion the query is actually circling. Everything in this table is derived from the vendor's own prices and the vendor's own 100-credits-per-minute rate, with the free 10 minutes included in the capacity.

Read the direction of that last column carefully, because it contradicts what every plan ladder implies. The cost per finished minute rises when you move up a tier. Starter is $1.29 a minute; Pro is $1.72. The reason has nothing to do with the credit price, which is proportional. It is the free 10 minutes: a fixed 10 minutes is a much larger share of a $29 subscription than of a $79 one. Per finished minute, Starter is the cheapest paid plan Synthesia sells. Annual billing moves both figures down, and because both tiers discount at the same fraction, the ranking between them never changes. Moving up a rung on this ladder buys features, seats and headroom. It has never bought a cheaper minute, at any volume, on either billing cycle.

That is the reframe worth carrying into your next budget meeting: on this product, the decision that saves you money is the billing toggle, not the tier.

The marginal minute has no published price at all

Now the finding that turns this from a table into a risk, and it is the one I would want a client to hear before signing anything.

Synthesia publishes no per-credit rate for extra minutes. Not a high one, not a tier-locked one. None. The vendor's own FAQ says video creation "pause[s] until your plan renews", and offers exactly two remedies: "you can buy extra credits from Plan & billing, or move to a plan with more credits". Its own announcement post describes the lower tiers' route as "upgrade your plan to unlock more credits". One third-party page states flatly that self-serve top-ups do not exist, which contradicts the FAQ, and I could not verify either claim on the vendor's account-billing article.

Either way, no per-credit price is published anywhere on the public web. Compare that with the vendor in the next seat on your budget: HeyGen publishes $0.05 per extra credit, Business-only. Synthesia publishes no equivalent number at all.

The practical consequence is blunt. A team whose month runs long cannot buy its way out at a known rate. It either pauses production or changes plan, and if it changes plan it is choosing between published tiers and unpublished credit multipliers. "Hidden costs" is the most repeated phrase in this query's third-party layer, and the word is wrong. The cost is not hidden, it is absent. You cannot audit a price that does not exist, which is why every explainer on this query eventually gives up and quotes the plan card instead.

One more thing to price before you commit, because it is your own waste rate and it is bigger than the credit rate. A re-render of changed footage bills again. The vendor's own worked example is a 2 minute 40 second video at 320 credits, revised by four seconds of new words, billing 324. That is a grain of sand. Now apply a real script-review cycle: a 10% re-render allowance takes the effective cost per finished minute from $1.29 to $1.42 on monthly Starter, and a 100% allowance, which is one unnecessary full take per kept minute, takes it to $2.58. I am giving you the formula and one labelled example rather than a universal waste percentage, because your review cycle is not mine. Plug yours in and the number moves.

Three team budgets, and the plan each one should buy

All three scenarios are derived, and the assumptions are the argument: typical video at 100 credits per finished minute, first-try renders, finished minutes equal generated minutes. Change the volumes to your own and the arithmetic still holds. Change the render assumption and every minute figure halves.

A 3-person content team shipping 30 finished minutes a month.

Read the first row twice. The "$29 tool" that every third-party page quotes cannot produce 30 minutes a month. Add any lip-synced dubbing and it gets worse, not better, because dubbing draws from the same pool at 80 credits a minute. The realistic entry point for a three-person team at this volume is the top self-serve tier, and on annual billing that is $588 a year, which is the number to put in the budget line.

A 10-person marketing department shipping 100 finished minutes a month. That workload is 10,000 credits.

At both of the first-party credit figures Synthesia publishes, no self-serve tier covers a 100-minute month. The capacity exists at the higher multipliers, and their prices are published nowhere, and the vendor's own plan finder declines to help ("Estimates only. Actual credit use depends on what's in your videos"). A ten-person department cannot price its own requirement from public information. That is not me being unable to find a number. It is a number the vendor has chosen not to publish, and the honest advice is to treat Enterprise as the tier this requirement lands on.

An agency producing for clients: five client videos and a month of localisation.

The largest credit consumer in an agency month is not video creation. It is lip-synced dubbing, at 4,800 credits against 2,000 for the source video, a 2.4x overhead, and every one of those minutes is a minute the "10 minutes included" headline never mentions. If you localise content for clients, you are not buying a video tool. You are buying a dubbing tool that also shoots video, and you should size it that way.

The three budgets together say something simple. Your tier is downstream of two things no ranking page names: the credit rate of the features you actually use, and your billing cycle. Ranked by leverage, the billing toggle is worth about 38%; the feature mix decides whether your month is priced like a video month or a dubbing month; and the tier buys features and headroom. Enterprise is the only rung with no ceiling, and the only one you cannot price.

Synthesia against the production routes it replaces

The comparison has to carry two scope warnings or it becomes vendor copy, and how metered AI pricing works is the same discipline applied to a different meter.

Warning one: those bands do not measure the same thing. The $2 to $30 band and the $13 to $220 band are both labelled "AI video generation", and they differ by an order of magnitude at the top. Anyone who quotes you whichever of them suits their argument is not comparing, they are selecting.

Warning two is the one that should actually change how you buy. A per-minute comparison against agency production is not like for like, because an agency minute includes strategy, scripting, shooting, editing and delivery, and Synthesia replaces only the shoot. The same literature makes the point in its own numbers: AI-assisted production moved the median corporate cost to roughly $2,500 per finished minute, not to $2. The same source set puts a 60-second video at about 27 minutes of production time with AI, against one to three weeks freelance. The saving is real and it is large, and it has not produced proportional returns, because the script, the review cycle, the re-render and the distribution all survive the switch. Synthesia at $0.80 a minute against a $2,500 median is not a 3,000x win. It is a cheaper shoot inside a workflow that still costs what it always cost.

Two context figures for sizing the bet. 41% of businesses now use AI for video production, up from 18% in 2023, and AI video generation volume grew 840% between January 2024 and January 2026. The AI video generator category is either $847M or $946M in 2026 depending on where the boundary is drawn, with the wider AI video marketing market at $18.6B and the US slice of the generator market at $182.8M. I am not picking one, because the sources genuinely disagree on scope. The vendor itself raised a $200M Series E at a $4B valuation in January 2026, led by Google Ventures, and claims 50,000+ business customers and 90% of the Fortune 100.

What the Basic plan covers, and where the pricing stops working

The free tier is real, and it is capped in four ways at once. First-party: "The Basic plan is free and includes 10 minutes of video, 3 survey responses and 500 credits every month, renewing each month", watermarked, one editor. The pricing card says 1,200 credits, the help centre says 500, and I cannot reconcile them, so plan on the lower one.

Three boundaries, each written as a decision rule rather than a complaint.

Keep draft work off the avatar. Consumption is footage-based, and a re-render of changed footage bills again. So write, review and rewrite with low avatar screen time, and commit avatar-heavy footage only once the script is locked. A screen-recording or stock-led cut at identical runtime costs a fraction of an avatar-led one. This is the single easiest saving on the platform and it is a discipline, not a plan upgrade.

Test whether you are buying minutes at all. On the card's figures, Basic and Starter share the same 10 free minutes, and a team producing under about 22 minutes a month that only needs logo removal and downloads is buying a feature list rather than capacity. Because the vendor's two pages disagree by 2.4x on the exact free allowance, treat that as conditional and name both numbers when you raise it. If your volume is genuinely tiny, the honest answer may be the free tier plus the $29 spent on the script.

If dubbing is your workload, stop reading the minutes column. At 80 credits per lip-synced minute against about 100 for video creation, a localisation-heavy month is priced like a video month while producing a fraction of the video content, at 2.4x the source video at three markets. No page in this field performs that comparison, and it is the most useful thing this article can hand an agency.

The bottom line: one recommendation per team size

One recommendation per shape, then the part about video that people selling video software rarely say.

• Solo or a three-person content team, under about 22 minutes a month, avatar-light: Starter on annual billing at $18 a month. Add dubbing and you leave this tier.
• Three to five people, 22 to 46 finished minutes a month: Pro billed annually, $49 a month and $588 a year. Not because the tier is cheaper per minute, because it is not, but because Starter cannot reach your volume and there is no top-up path.
• A ten-person department at 100 minutes a month: you have left self-serve. Either buy a Pro credit multiplier and accept that its price is unpublished, or open the Enterprise conversation, and go in knowing that no public number exists for either.
• An agency localising client work: size the plan on dubbing, not on video. Lip-synced dubbing runs 2.4x the credit cost of the source video at three markets, and it draws from the same pool.
• Anyone producing under 10 minutes a month with a watermark acceptable: stay on Basic and spend the $29 on the script.

The last thing I would say to my own team before we added a video line. Video is one channel in a campaign, not a separate vendor relationship. The script, the review cycle, the localisation, the ad variants and the reporting are the same work whichever tool renders the frames, and the $2,500 median above says the overhead travels with the workflow rather than with the render. That is the argument for running video inside the platform that already holds your campaigns, not beside it. Allable covers that chain from one conversation, with the campaign, SEO, social and asset work sitting in the same place as the video brief, on a 7-day trial, and where a video tool sits in the stack is a budget question before it is a tooling one. Cost the video work as one line in a stack you already pay for rather than as another subscription with its own renewal date, and see whether the arithmetic survives. It is the same discipline that makes seat-based plans and another tool that renamed its plans worth reading before you commit a year, and the same one that makes a per-credit rate worth demanding from any vendor whose meter you cannot see.