HeyGen V5 is 8 credits a second

HeyGen Avatar V5 on Versely is 8 credits a second. The floor is 40 credits. The ceiling is 960. Billing is duration-driven. You do not upload a still to this row. This page is that lipsync meter. HeyGen's dollar plans are a different invoice, and they already have a page: HeyGen avatar price 2026. Quote this row in catalog credits. Quote that page in HeyGen's own units. Do not invent an exchange rate between them.

The published slug is HeyGen Avatar V5 digital twin. Content type is lipsync. Category is premade-avatars. Audio is on. An image is not required. Output qualities on the card are 720p, 1080p, and 4K. Aspects are 16:9, 5:4, 1:1, 4:5, 9:16, and auto. The duration list is empty. Length follows the take. The catalog then charges 8 credits for every billed second, never less than 40, never more than 960 on this row.

Eight credits a second is the live meter

Eight is the per-second cell. It is full price and discounted price. There is no percent-off on this row. If a spreadsheet still says 10 credits a second, or a 50-credit sticker, that spreadsheet is reading an older snapshot. The live catalog card on 19 September 2026 lists per-second billing, duration-driven, 8 credits full, 8 credits discounted.

Forty on the card is the display floor, not a flat film. A producer who budgets "one V5 video equals 40 credits" is quoting the shortest billed take, then shipping a 45-second explainer and wondering why the pool moved. Forty is what you pay when the take is at or under the floor. Everything longer is 8 times the billed seconds.

Speech is the product. Audio is on because the take is a mouth on a track: script in, or audio in, talking twin out. Silence is not a personality. If you do not have the line yet, you do not have this job. Write the line. Then open the row.

Premade-avatars means you pick talent the catalog already has, or a twin you already enrolled, not a paragraph that invents a new face in a new room. An image is not required because the plate is the avatar, not a JPEG you designed this morning. If the brief is "make this founder headshot talk," you are on a talking-photo job, not this meter. That split is the whole point of Talking-photo is not a UGC avatar.

4K is on this row. The V3 sibling does not list it. If delivery is a 4K presenter master, that is a reason to pay 8 a second instead of 3. If delivery is a 9:16 talking close-up that will be recaptioned and compressed, 4K is optional. Do not spend the V5 meter as a habit because the name is newer.

On this snapshot the price matrix has one cell: 8 credits a second. Length is the dimension that moves the bill. Aspect does not. A quality toggle is not published as a second price band here. Confirm at generate if that ever splits. Do not budget a 4K multiplier this card does not show.

Auto aspect is on the list. Auto is not a creative direction. Pick 9:16 for Stories and Reels, 16:9 for YouTube, 1:1 when the placement is square. Generating auto and cropping a face you paid 8 a second to sync is how you throw away the sides of the mouth. 5:4 and 4:5 are here for placements that are neither wide nor vertical. Use them when the slot is that shape. Do not force 16:9 and pad.

This is still a lipsync row, not a text-to-video scene. It will not build a warehouse, a handheld UGC kitchen, or a pack orbit. It will put words on a twin. Pay it for that. Skip it for the rest.

Forty is the floor, 960 is the ceiling

Divide the floor by the rate and you get five billed seconds. Divide the ceiling by the rate and you get 120 billed seconds. That arithmetic is not a duration menu. The slug does not sell 5s / 10s / 15s as named options. You get a take. The catalog multiplies. If the take would have been cheaper than 40 at 8 a second, you still pay 40. If the take would have been dearer than 960, you are off this row's published range.

Worked credits at 8 a second, live catalog 19 September 2026:

• At or under five billed seconds: 40 (the floor). You do not get a 16-credit three-second bargain on this row.
• 8 seconds: 64
• 15 seconds: 120
• 30 seconds: 240
• 45 seconds: 360
• 60 seconds: 480
• 90 seconds: 720
• 120 seconds: 960 (the ceiling)

Retries are a second line. Three 15-second hook variants are 360 credits, not 120. A 30-second explainer plus one safety recut is 480, not 240. Put the retry cap on the brief before anyone presses generate. The floor makes short misses expensive relative to their runtime. The per-second cell makes long misses expensive in a more obvious way. Both are still this meter. Neither is a pack price in dollars.

Cut the script before you generate. Throat-clearing, a ten-second hello, and a recap that repeats the hook are billed seconds with no extra information. The cheap edit is the one you do in the doc. The dear edit is the one you do by regenerating a 60-second twin because the third sentence was wrong.

960 is a cap, not a dare. A two-minute twin read is a course module shape, not a social hook. If the week is ten 15-second ads, budget 120 each plus retries, not one 960 dump. If the week is one long presenter, confirm the take still fits under 120 billed seconds on this row. Do not assume a 15-minute lecture belongs here.

A 30-second talking twin is 240 credits on V5. That number is only the mouth. If the finished ad also needs a still, a motion plate, a voiceover on a different meter, captions, and an export, those are other rows. Do not quote 240 as "the ad." Quote 240 as this lipsync line, then add whatever else the brief actually runs.

Do not convert any of those credit numbers into dollars. Versely credits are catalog credits. Pack prices on an account are a different screen. Third-party USD belongs to the third party. A homemade credits-to-dollars rate is how a margin you do not actually know ends up in a client email.

V3 is 3 credits a second, not a 17-credit sticker

The sibling row is HeyGen Avatar V3 digital twin. Same job class on paper: lipsync, premade-avatars, audio on, image not required, duration-driven, empty duration list. The live meter is 3 credits a second. The floor is 14 credits. The ceiling is 327. Qualities run 360p through 1080p. There is no 4K token on that card.

Older mouth-pass posts locked V3 at a 17-credit floor. That snapshot is stale. Do not quote 17 in a live estimate. Quote 3 a second, 14 minimum, 327 maximum, and re-read the card on the day you generate. The same warning applies to V5 posts that still say 50 credits or 10 credits a second. Those were earlier matrices. Eight, 40, and 960 are the live V5 numbers.

Same 30-second read: V5 is 240 credits, V3 is 90. Same 15-second hook: V5 is 120, V3 is 45, unless V3's 14-credit floor is still the larger number on a very short take. The 8 versus 3 gap is the whole premium. You are paying it for the V5 twin, the 4K option, and the higher ceiling, not for a different billing type. Both rows multiply by duration. One multiplies harder.

Pick V3 when a library presenter at 3 a second is the talent you actually wanted, and 1080p is enough. Pick V5 when the brief named this engine, or 4K, or the longer ceiling. Do not pick V5 because the version number is bigger. Do not pick V3 because an old blog said 17 and 17 looked cheap next to a cinematic generate. Seventeen is not the live floor. Three a second is.

Do not average the two rates. A week that needs four V5 takes and six V3 takes is four times the V5 meter plus six times the V3 meter. Mixing them in one cell called "HeyGen" is how a quote lies. Name the slug. Name the seconds. Multiply.

V3's 327 ceiling is lower than V5's 960. A long module that fits V5 may not fit V3. Check the ceiling before you write a four-minute script onto the cheaper row and discover the catalog will not take it. Checking is cheaper than rewriting the module after a failed generate.

Iterate on V3 only if V3 is the talent you will ship. Do not draft on V3 at 3 a second and then "upgrade" the winner to V5 and expect the same face, the same 4K, and the same mouth. Different rows. Different twins. A cheap draft on the wrong identity is not a saving. It is a reshoot.

Skip this meter when the shot is not a twin

Eight credits a second is the right bill for a talking twin on this catalog. It is the wrong bill for three other jobs that still get filed under "avatar."

Handheld UGC is a body, a room, and a product in a hand. That is not a desk twin. Run it through the UGC video generator, not through Avatar V5. Paying the lipsync meter to get a polished presenter, then wondering why it does not look like a creator kitchen, is a category error. The twin is doing what it is for. The brief was not.

A talking still from a photo is talking-photo lipsync. Different input, different product. The map for that split is Talking-photo is not a UGC avatar. V5 does not require an image. If you are clutching a headshot, you are not on this row's happy path. Do not enroll a twin for a one-still ad you will never repeat.

Shopping HeyGen as a monthly seat is the other invoice. That page already exists. This page will not retell plan stickers, seat add-ons, or per-minute burns inside HeyGen's own product. If the team already lives in that editor and wants that editor's minutes, use HeyGen avatar price 2026 and recheck HeyGen's own table on the day you quote. If the twin is one named row inside a week that also needs captions, silent motion, and a UGC body, stay on this meter and sign up at https://app.versely.studio/signup.

Captions still sit after the take. A talking mouth is not a transcript. Mute autoplay hides the work you just paid 8 a second for. Burn the line once the take is the take. Do not ask V5 to typeset. Do not ask V5 to score the clip. Speech is the output. Type and music are later passes.

The test is boring, which is why it works. If the file you will ship is a named twin saying an approved script, pay 8 a second and cut the script. If the file is a still, a handheld body, or a seat in someone else's app, this meter is not the quote. Read the live card. Ignore the cached 17. Ignore the cached 50. Eight, 40, 960.

FAQ

How many Versely credits is HeyGen Avatar V5?

On the 19 September 2026 catalog, HeyGen Avatar V5 digital twin is 8 credits a second, minimum 40, maximum 960, duration-driven, no image required. A 15-second take is 120 credits. A 30-second take is 240. Shorter than five billed seconds still costs 40. HeyGen Avatar V3 digital twin is the 3-credits-a-second sibling, floor 14, ceiling 327.

Why do older pages say 50 credits or 17 credits?

Those were earlier snapshots. V5 mouth-pass posts listed a 50-credit floor and, in one matrix, 10 credits a second. V3 posts listed 17 as a floor. The live catalog on 19 September 2026 is V5 at 8 a second (40 to 960) and V3 at 3 a second (14 to 327). Trust the card you confirm at generate, not a cached 17.

Is this the same as paying HeyGen's monthly plans?

No. This is Versely's lipsync meter on a published catalog row. HeyGen's own plans are a subscription plus a credit burn inside that product. They are documented on HeyGen avatar price 2026. There is no honest "N Versely credits equals one HeyGen minute" line. Pick one invoice and stay on it.