Higgsfield can carry an AI influencer from first face to talking clip inside one browser account. AI Influencer Studio builds the character from menus for 1.12 credits a generation, Soul ID locks a face trained on 20 to 80 photos, and Lipsync Studio or Intone gives the persona a voice. The realistic plan for an account that posts video is Plus, at $59 a month or $47 a month billed yearly, because Starter at $19 a month gets one lipsync model and no Seedance 2.5. One rule outranks every tool. Any real face or voice you feed in needs that person's consent and release [1].

The Toolkit take
1.12 credits
per AI Influencer character generation (close-up plus full-body shot), cost shown before you confirm
20-80 photos
train one Soul ID, one person per ID, not exportable
3-30 s
AI Influencer motion reference clip, output at 1080p
up to 30 s
of speech per Intone generation, at 720p or 1080p

Designing the persona in AI Influencer Studio

AI Influencer was rebuilt in October 2026 as a menu-based builder, so you shape a character without writing a prompt. The character types run from humans to mammals, reptiles, fish, hybrids and aliens, and the output reaches 4K. Higgsfield's October 2026 post describes 19 menu settings, each tuned on a three-level scale from Average to Extreme. Prompting survives only as an optional last step for details the menus do not cover. That suits a social manager who knows the look they want but does not want to learn prompt syntax.

Each generation returns a close-up portrait and a full-body shot for 1.12 credits, and the cost appears before you confirm [2]. Text edits to a finished character also cost credits, shown on the Edit button. Every user got 5 free AI Influencer generations in October 2026; the pricing page lists 10 on Starter and 30 on Plus or Ultra. You may upload your own face to start from. Use only your own photo or one from someone who gave permission. Invented faces avoid that question entirely.

Locking one face with Soul ID

A menu-built character can drift between sessions. Soul ID is the fix. It trains on 20 to 80 photos of one person, takes a few minutes, and holds exactly one identity per ID [3]. For a synthetic influencer, the help center says to train it on 20 or more portraits of the character itself, so your AI Influencer outputs become the training set. The trained ID works in Soul 2.0 and Soul Cinema and carries into Seedance automatically as an Element. Higgsfield's posts put the training cost at 25 credits, paid once.

The pricing page caps Soul ID generations by plan: 108 on Starter, 480 on Plus and 1,200 on Ultra in its yearly view, or 9, 40 and 100 in the monthly view, with Free allowed up to 1. The ID cannot be exported as a file. It lives in the account that trained it, and credits and generations cannot move between accounts. The account email cannot be changed either. Open the account under the business address that will own the persona, not a freelancer's personal login.

💡
Toolkit tip
Keep test looks private

Higgsfield generations are private by default and appear in Explore only if you publish them or turn auto-publishing on, so leave auto-publishing off until launch.

Curious how Higgsfield feels in practice?Try Higgsfield →

Filling the feed: Soul 2.0, Soul Cinema and Popcorn

Most influencer posting is stills. Soul 2.0 is Higgsfield's own image model, with output up to 2K, more than 20 curated Moodboards since February 2026 and Soul HEX for color control, at 0.12 credits per image. Soul Cinema produces cinematic stills without presets or Moodboards but still accepts Soul ID and Soul HEX. Paid plans also come with a pool of free Soul 2.0 and Soul Cinema generations: 300 on Starter, 5,000 on Plus and 10,000 on Ultra in the yearly view. That pool is granted once at purchase. It does not renew, and it works on higgsfield.ai only.

Popcorn handles the carousel. It builds coherent sequences of up to 8 frames from up to 4 image references, in Manual, Auto or Instadump mode, at 1.5 credits per image. Instadump packs can start from up to 15 reference photos, which fits the casual dump format many lifestyle accounts post. A Higgsfield post adds Soul Photodump, a set of 15 photos of one Soul ID character in different settings. Same face, new outfit, new location. That is the whole job of a lifestyle feed.

💡
Toolkit tip
Train Soul ID before you post

Generate the look in AI Influencer, then train a Soul ID on 20 or more portraits of that character so the face holds across Soul 2.0, Soul Cinema and Seedance.

Motion and voice: Genjutsu, Lipsync Studio and Intone

Every AI Influencer video runs on Genjutsu motion transfer. You upload a reference clip of 3 to 30 seconds, and its movement, camera and timing are copied frame for frame onto your character, with up to 40 images per video and 1080p output. The Motion tab also offers Objects swap at 480p, 720p or 1080p. The help center's alternative is Motion Control with the character image and a video you recorded yourself; Kling 3.0 Motion Control starts at Plus. Finished characters also appear in Cinema Studio through a dedicated tab. Our Genjutsu guide covers its modes and limits.

Speech comes from Lipsync Studio, which pairs an image or video with uploaded or generated audio and then a model; the help center lists eight. Starter gets Wan 2.5 Speak only, while Higgsfield Speak 2.0, the Kling lipsync models, Infinite Talk and Sync Lipsync 3 start at Plus. Speak 2.0 costs about 14 credits per 5 seconds at 720p, and Sync Lipsync 3 about 18 credits per 2 seconds at 4K. Intone, added in October 2026, makes lip-synced talking characters with up to 30 seconds of speech per generation at 720p or 1080p.

The voice itself can come from Higgsfield Audio, which offers text to speech, voice change, translation into 18 languages and cloning from an MP3 or WAV sample. Cloning is for consented voices only. A creator cloning their own voice for a branded twin is fine. A celebrity sound-alike is not. Seedance 2.5, on Plus and up, can also generate audio in the same pass as the clip, with your Soul ID character attached as an Element.

⚠️
Watch out
Starter talks with one model

On the Starter plan, Lipsync Studio offers Wan 2.5 Speak only; Higgsfield Speak 2.0, the Kling lipsync models, Infinite Talk and Sync Lipsync 3 start at Plus.

What a posting schedule costs on each plan

Plan Price Credits a month Free AI Influencer generations Persona limits
Free $0 None 5 for every user (October 2026) Visible watermark on every output; no commercial use
Starter $19 a month, monthly or billed yearly 270 10 One lipsync model (Wan 2.5 Speak); selected video models only; no Unlimited window
Plus $59 a month, or $47 a month billed yearly 1,200 30 All lipsync models; 7-day Unlimited window on named models, higgsfield.ai only, standard queue
Ultra $129 a month, or $99 a month billed yearly 3,000 30 All lipsync models; 7-day Unlimited window on named models, higgsfield.ai only, standard queue

The upgrade point is concrete. By our arithmetic, Plus at $59 a month plus a $90 Auto-Refill comes to $149 for 2,820 credits, while Ultra at $129 a month carries 3,000 credits. Once you would refill Plus by $70 or more a month, Ultra gives more for less. Billing yearly, Plus costs $564 a year against $708 at the monthly price, a $144 saving that pays off only if you stay 10 months or more. Subscription credits never roll over, as our pricing breakdown explains.

Video is where the credits go. By our arithmetic, at the posted rate of 105 credits for a 15-second Genjutsu Motion Transfer clip at 720p, the 1,200 monthly credits on Plus cover about 11 clips and the 3,000 on Ultra about 28. A 5-second Seedance 2.5 clip at 720p costs 35 credits at the posted rate, so Plus covers about 34 and Ultra about 85. Those counts assume one take each. Higgsfield's own post says three to five generations per usable clip is normal, so a daily-video persona outgrows Plus quickly.

Unlimited windows will not carry a scheduled posting machine. They cover named models for a set number of days, on higgsfield.ai only, in the slower standard queue, and for personal human use; automation, scripting and credential sharing are banned and can pause the window for review. Anything run through MCP, the CLI or Canvas spends plan credits.

⚠️
Watch out
Free-plan outputs stay watermarked

Every Free-plan generation carries a visible Higgsfield watermark, upgrading does not clean earlier outputs, and commercial use comes with a paid plan.

Ready to put Higgsfield to the test?Try Higgsfield →

Higgsfield's Terms make you responsible for holding rights, releases and consents from anyone whose name, likeness or voice appears in your inputs, and face or voice inputs carry a separate warranty of consent. You may not use the service to identify or verify people, or submit biometric templates. Higgsfield says it uses face and voice data transiently and does not store or sell it. Reported intimate images shared without consent are removed within 48 hours. If a human model's photos train your Soul ID, get a signed release first. Our face swap guide covers the same rules for swaps.

Disclosure is a legal duty where the law requires it: you must label AI output and must not strip the provenance markings Higgsfield applies. Those may include invisible watermarks, metadata or C2PA Content Credentials, though Higgsfield does not promise every output keeps them. A paid plan removes the visible watermark, and commercial use comes with a paid plan. Higgsfield claims no ownership of your inputs or outputs, but outputs may not be unique. One more caveat. Higgsfield may train its models on your content unless you are on Enterprise, and deleting content stops only future use.

Frequently asked questions

Is there a Higgsfield app for posting from my phone?+

No. Higgsfield has no mobile or desktop app and runs in the browser on any device; a downloadable app calling itself Higgsfield is a copy.

Do free-plan influencer images lose the watermark if I upgrade later?+

No. Upgrading to a paid plan does not remove the watermark from earlier Free-plan generations, so you regenerate them on the paid plan.

Can I build the influencer from inside Claude?+

Yes, on a paid Higgsfield plan: the MCP connector reaches Soul characters, Elements, voiceover and voice clones (5 per workspace by default), and every generation spends plan credits at standard rates, with no Unlimited.

Can I use a voice clone of a real person for my persona?+

Only with that person's consent: Higgsfield Audio clones a voice from an MP3 or WAV sample for consented voices only, and the Terms require releases from anyone whose voice is in your inputs.

The bottom line

Higgsfield covers the full persona loop in one account: a menu-built character, a Soul ID that holds the face, Soul 2.0 and Popcorn for the feed, Genjutsu motion, and lipsync or Intone for speech. Starter suits a stills-first account testing a look. A persona that talks on video belongs on Plus, and moves to Ultra once refills pass $70 a month. Keep the account under your business, keep releases on file, and label AI output wherever the law asks.

Sources
  1. [1] Higgsfield Terms of Use Agreement (2026-11) https://higgsfield.ai/terms-of-use-agreement
  2. [2] Higgsfield Help Center: how do i use ai influencer (2026-11) https://higgsfield.ai/creator-hub/help-center/tools/how-do-i-use-ai-influencer
  3. [3] Higgsfield Help Center: how do i create and use a soul id character (2026-11) https://higgsfield.ai/creator-hub/help-center/ai-models/how-do-i-create-and-use-a-soul-id-character
  4. [4] Higgsfield pricing (US) https://higgsfield.ai/pricing
All 6 sources we used for this article
  1. 1. Higgsfield Terms of Use Agreement (2026-11) https://higgsfield.ai/terms-of-use-agreement
  2. 2. Higgsfield Help Center: how do i use ai influencer (2026-11) https://higgsfield.ai/creator-hub/help-center/tools/how-do-i-use-ai-influencer
  3. 3. Higgsfield Help Center: how do i create and use a soul id character (2026-11) https://higgsfield.ai/creator-hub/help-center/ai-models/how-do-i-create-and-use-a-soul-id-character
  4. 4. Higgsfield pricing (US) https://higgsfield.ai/pricing
  5. 5. Higgsfield Help Center: how credits work https://higgsfield.ai/creator-hub/help-center/credits/how-credits-work
  6. 6. Higgsfield changelog https://higgsfield.ai/creator-hub/changelog