AI video for a faceless channel: the credit math on shipping one clip a week

Every "best faceless AI YouTube channel" article in the SERP will tell you three true things. The CPM order is roughly finance, AI/tech, business, real estate, then everything else. The format that beats the stock-footage-plus-robotic-voice template is a channel with original visuals and a recurring character. And starting is a four-step loop: pick a niche, lock a character, ship a cadence, iterate on retention.

None of them will tell you what it costs. A faceless channel that posts one clip a week on an AI studio should be budgetable in credits, in advance, before you commit to the format. Here is that number, broken down the way I actually plan it.

The format that works, in one paragraph

A locked character — animated, 3D or stylised — talking to camera about one topic, 60 to 90 seconds, 9:16 vertical for TikTok and Reels, voiced, native audio where you can get it, with cuts that keep the viewer's attention past the three-second drop. One recurring character across every clip is the thing a faceless channel can't fake with stock footage; it is also the thing that compounds, because a viewer who recognises the character on scroll stops longer than a viewer looking at generic b-roll.

The engines our shelf serves are the engines a faceless channel runs on. Kling O3 Pro for the character scenes with camera moves, 450 credits per 15 seconds voiced at 720p. Veo 3.1 Fast for 8-second openers and short reactions, 240 credits voiced or 160 silent. Seedance 2.5 for the one signature shot a week where the whole clip hangs on one unbroken 15-second take, 1,350 credits.

One clip a week, priced before you commit

Four clips a month. One locked character. Here is what a month of posting costs on three realistic cadences. Numbers from the engine picker, not a provider list rate — these are the credits the shelf prints before you press render.

The lean cadence. Four 60-second clips, each four 15-second Kling scenes with one 8-second Veo opener, one character sheet, one writer-driven script per week, one export per clip.

step, per week	credits
writer, five actions	25
character sheet (one-off, first week only)	20
opener on Veo 3.1 Fast, 8s voiced	240
three 15s Kling scenes voiced	1,350
one retake per clip (realistic)	about 450
export	20
weekly total	about 2,085

Four weeks at ~2,085 is about 8,340 credits a month. On Creator at $45 for 6,000 credits plus one $25 top-up (2,750 credits) you clear it with change: total $70, which is about $17.50 per 60-second finished clip. On Studio at $99 for 15,000 credits you have 7,000 credits of headroom to retake and experiment.

The standard cadence. Four 90-second clips, same mix with one Seedance 2.5 signature shot per clip for the beat you want people to clip.

step, per week	credits
writer, five actions	25
opener on Veo 3.1 Fast, 8s voiced	240
four 15s Kling scenes voiced	1,800
one Seedance 2.5 15s signature at 720p	1,350
retakes, 30% of the hero/signature	about 950
export	20
weekly total	about 4,385

Four weeks is around 17,540 credits — about $132 at Lite's one-cent rate, about $99 at Creator's. The Creator plan ships you exactly 24,000 credits a year on a monthly cycle; the standard cadence spends 70% of that on content and leaves 30% for retakes, experiments and the week you want to post twice. Studio at $99 a month and 15,000 credits covers it with margin.

The ambitious cadence. Daily clips, 60 seconds, one character, no signature shot (the shot that earns the shelf's premium needs a week to breathe, not a day).

A daily 60-second clip at the lean rate is around 2,085 credits a week per clip × seven = about 14,600 credits a week. That is 58,000 credits a month, which is the Studio plan plus two top-ups or an Agency plan. Realistic if the channel is working; expensive if you're still finding the format. The honest teaching: start on the lean cadence, scale when retention is above 40% on a 15-second mark.

Which niche this actually works for

Finance, AI tool reviews, business history, software demos, product comparisons, calm sleep narration, dark history, true crime — the SERP has the niche list right. What the SERP doesn't rank is which of those niches most reward a locked character.

Niches where a recurring character compounds hard. A mascot in a business history channel. A single narrator-persona in dark-history. A character who tries each AI tool in an AI-review channel. A fitness coach-persona who demos one exercise a day. A real-estate agent-persona who tours one property a week. The character is the brand before the content is.

Niches where a locked character hurts. Pure sleep music, pure meditation, calm nature. The format works better on b-roll and voice; a character on screen pulls focus away from the function.

The highest-CPM niches on a locked-character channel. Finance and legal/tax education carry $15 to $45 CPMs per the current data (Blotato, outlierkit, 2026-10). A finance channel with a locked analyst-character is the single-highest expected value per view a faceless channel can run today. B2B SaaS at $14 to $35 CPM comes second and is the easier format for an AI studio to shoot: product scenes, locked presenter, demo cutaways.

The three mistakes most faceless creators make with a locked character

They render too hot. Three 15-second hero shots on Seedance 2.5 is 4,050 credits per clip — $40 on Lite. Done weekly, that's $160 a month on signature-engine renders for 240 seconds of finished video. The draft pass is a 480p Seedance 2.0 Fast render at 25 credits a second — a fifteen-second shot is 375 credits. Draft, decide, then render. Skip it and you'll pay the premium for a storyboard you'd have torn up.

They don't lock the prompt. The character sheet is one artefact. The descriptor block for that character — the hair, the eyes, the one piece of clothing people recognise — has to be the same wording in every scene's prompt, with only the action and the camera changing. This is the single piece of advice every character-consistency article in the SERP converges on, and it is true.

They skip the opener. The one place Veo 3.1 Fast earns its share on a faceless channel is the opener. Eight seconds, locked camera, native audio, one spoken line into camera. 240 credits for a hook that catches the three-second drop. On Kling the same shot is slightly cheaper per second but doesn't have native lip sync; on Seedance 2.5 it is more than three times the price. Veo wins exactly this one job.

What a free account actually buys you

300 welcome credits are enough for one character sheet (20), one Veo 3.1 Fast 8-second silent test (160), and about 120 credits of writer runs (24 actions) before you spend a dollar. The sheet is the asset you keep; the test is the format proof. If both land, upgrade to Lite at $10 a month (1,000 credits) and ship your first clip inside a week.

The jump from Lite to Creator is where most faceless channels find their legs. $45 for 6,000 credits is three to four 90-second clips a month at the standard cadence, with headroom for retakes and experiments. From Creator you can tell whether the format is working — retention at the 15-second mark, impressions per week, click-through on the handle — before you commit to Studio at $99.

The honest part

Our own 9.7 million views across 30 days on @cevrico came from running a character-locked faceless channel on exactly this chain. One character, mixed engines, draft-hero-signature pass on every clip. Own results, not a projection; what your channel does will depend on what you shoot and the niche you shoot it in. The credit math above is specific because the engine picker in the studio prints every number before you press render.

Pick a niche and ship a draft this week. The format compounds; the first clip is the hardest and the cheapest.