HeyGen

Learning Objectives

• Understand what HeyGen is and how it differs from text-to-video generation models
• Learn HeyGen's core capabilities: AI avatars, video translation, and avatar cloning
• Identify the business and creator use cases where HeyGen delivers the most practical value

What Is HeyGen?

HeyGen is an AI video platform founded in 2020 (originally D-ID competitor, launched as HeyGen in 2022) that has become the dominant tool for AI avatar video creation. Rather than generating cinematic footage from text prompts, HeyGen specializes in talking-head videos — professional presenter-style content where an AI avatar reads a script, explains a concept, or delivers a message.

HeyGen's primary audience is businesses: sales teams, training and enablement departments, marketing teams, HR, and customer success organizations that need professional explainer videos at scale without the cost and time of traditional video production.

Pricing

Credits, not minutes, are the currency — and the rate depends on which avatar model you pick, which is the single most important thing to understand about the pricing. Avatar III costs roughly 3 credits per minute; the newer Avatar IV and Avatar V cost about 20 credits per minute. Video translation with lip-sync runs about 5 credits per minute. So a Creator plan's 600 credits stretches to a few hours of Avatar III output or well under an hour on the newest model.

Two things commonly misread: a custom digital twin is available on the free tier (one of them), not gated behind Business as it once was; and API usage is billed separately from the web subscription, so an API-driven workflow does not draw down your plan credits. Business adds seats at $20 per additional seat.

Core Capabilities

AI Avatar Library

HeyGen provides a library of more than 1,100 pre-built AI avatars — diverse in appearance, age, ethnicity, and presentation style. Each avatar reads your script and produces a video with synchronized lip movement, natural gestures, and realistic eye contact. No camera, studio, or editing required.

Alongside the stock library there are two ways to make an avatar of a specific person: a photo avatar, built from a single front-facing headshot, and a custom avatar, built from a short recording of the person themselves.

Avatar generations: III, IV, and V

The avatar model you choose changes both quality and cost, so it is worth knowing which is which.

• Avatar III is the established workhorse and by far the cheapest to run at roughly 3 credits per minute. It is the right default for high-volume, straightforward presenter video.
• Avatar IV takes a photo and a script and produces expressive facial movement and hand gestures. It is also available through the API, which is what lets developers embed avatar generation directly in a product.
• Avatar V, the fifth-generation model and the newest, is the one that changed the trade-off. Earlier you had to choose between the flexibility of a photo avatar and the realism of a video avatar; Avatar V does both. It captures your movement patterns from a 15-second video, then applies that motion to any photo you supply as the base look. It supports videos up to three minutes.

Video Translation & Dubbing

HeyGen's video translation feature is one of its most distinctive capabilities. Upload any existing video with a speaker and HeyGen will:

1. Transcribe the audio
2. Translate the text to the target language
3. Re-dub the speaker's voice in the target language
4. Remap the speaker's lip movements to match the translated speech

Translation now covers more than 175 languages, and it preserves the speaker's own voice through voice cloning rather than substituting a generic one. On paid plans a single video can be translated into up to ten languages at once. The free tier allows three translated videos per month, each up to three minutes.

Custom Avatar Cloning

Creating a custom avatar — a digital twin of a real person — no longer requires an enterprise contract. The free tier includes one, Business includes five, and Enterprise ten or more. The recording burden has also collapsed: a custom video avatar needs a short clip, as little as 30 seconds, and Avatar V needs only 15 seconds of motion reference plus a base photo.

HeyGen requires verified permission to build a custom avatar, and using one to impersonate someone is prohibited by its terms. That consent step is the meaningful control here, and it is worth reading before you plan a rollout.

This is transformative for organizations that want a branded, consistent human presenter across all videos without requiring that person to record every new piece of content.

Scenes, Highlights, and Developer Tooling

HeyGen has expanded well past single talking heads:

• Seedance 2.0 brings cinematic motion and camera work with identity-verified faces, and supports up to three avatars in a single scene.
• Instant Highlights v2 turns long-form recordings — podcasts, webinars, livestreams — into publish-ready short clips with facial tracking, dynamic framing, and automatic translation.
• HyperFrames is an open-source, HTML-based video framework under Apache 2.0 that makes video composition a code artifact, so it can carry version control, code review and reproducible builds.
• HeyGen CLI wraps the version-three API for command-line generation with structured output, and HeyGen Skills brings avatar generation into Claude Code and Cursor.

Script-to-Video

Type or paste a script (or have HeyGen's AI generate one) → select an avatar and voice → choose a background → generate the video. The entire production process takes minutes rather than days.

Talking Photo

Animate a single still photograph into a brief talking-head clip — useful for social media content, introduction videos, and rapid personalized outreach.

Strengths

• Fastest path to professional talking-head video — no camera, no studio, no editing software required
• Video translation with lip-sync — the most practical solution for distributing video content across languages at scale
• Custom avatar cloning — personalized digital presenter maintains brand consistency without recording sessions
• Enterprise-ready — SOC 2 compliance, SSO, data processing agreements available; widely used by Fortune 500 teams
• Ease of use — browser-based, minimal learning curve; teams can produce polished video without video production experience

Limitations & Considerations

• Not for cinematic or abstract video — HeyGen produces talking-head content, not generative scene footage; it is complementary to tools like Sora and Runway, not a replacement
• Avatar quality variability — stock avatars are good but custom avatars require a clean, high-quality recording source to produce convincing results
• Credit budgeting is not obvious — the newest avatar models cost roughly seven times more per minute than Avatar III, so a plan that looks generous in credits can be thin in practice if you default to Avatar V; check the rate before committing to a model for a large library
• Uncanny valley risk — AI avatars are convincing for business video but can appear slightly artificial in close-up, especially with unusual expressions or complex emotion
• Privacy: Custom avatar creation is subject to HeyGen's consent verification process; misuse for impersonation is prohibited by ToS

Best Use Cases

A named deployment worth studying

Harvard Business School uses HeyGen-built avatars of its own instructors inside HBS Foundry, an eight-week program for entrepreneurs priced at $699. The avatars give feedback during practice pitches and mock board meetings, alongside weekly live sessions with the real instructors.

Two details make it instructive rather than promotional. The program's director has said the avatars replaced an earlier plain-chatbot design because students asked for something more guided — so the avatar answered a teaching problem rather than being the starting point. And one instructor called his own digital copy "creepy" while reporting that his students liked it, which is a fair summary of where this technology currently sits: convincing enough to be useful, still recognizably synthetic. A reporter who tried the program noted that the avatar's smile stayed frozen throughout her pitch.

The transferable lesson is that the avatar worked here as a practice partner, giving low-stakes reps against something responsive, with the real humans kept for the live sessions. That division of labor is a better template than replacing an instructor outright.

When to choose alternatives:

• Cinematic scene generation → Sora 2 or Veo 3
• Creative short-form social video → Runway ML or Pika Labs
• AI video editing (cutting, overdub) → Descript
• Free or low-cost AI video experimentation → Synthesia free tier or Pika Labs

Getting Started

1. Go to heygen.com and create a free account
2. Click Create Video → choose AI Avatar Video
3. Select an avatar from the library, or build your own — the free tier includes one custom digital twin
4. Enter or paste your script in the text panel — adjust the voice tone and pacing
5. Choose a background (template, solid color, or custom upload) and set the aspect ratio
6. Click Generate — your video is ready in minutes; download in MP4

Key Takeaways

• HeyGen is the leading platform for AI avatar video — the fastest way to produce professional talking-head content without cameras, studios, or editing
• Video translation with lip-sync is a standout differentiator: distribute a single video to more than 175 language markets in hours rather than weeks, in the speaker's own cloned voice
• Pricing is denominated in credits, not minutes, and the avatar generation you choose drives the burn rate — Avatar III is roughly 3 credits per minute against about 20 for Avatar IV and Avatar V
• Custom digital twins are no longer an enterprise-only feature: the free tier includes one, and Avatar V needs just a 15-second motion clip plus a base photo
• Harvard Business School's use of instructor avatars as pitch-practice partners inside a paid program is a useful template — the avatar supplements the live instructor rather than replacing them
• HeyGen is not a replacement for cinematic or abstract video tools — it excels specifically in the business presenter format; pair it with Sora, Veo 3, or Runway for full video production capability