HeyGen

HeyGen is a Los Angeles-based AI video company best known for avatar video and video translation: a generated or cloned presenter speaking a script, or real footage re-voiced and lip-synced into another language. Founded in 2020, it built its audience among marketers and creators, and with Synthesia it is one of the two largest companies in AI avatar video. It reported $200 million in annual recurring revenue in June 2026. As of October 2026 its newest avatar model is Avatar V, and its Video Agent can produce videos up to 30 minutes long.

Company

HeyGen was founded in 2020 by Joshua Xu and Wayne Liang. In June 2024 it raised a $60 million round led by Benchmark that valued it at about $500 million, and its growth since has been among the fastest in generative video. It claims a large enterprise footprint, though its own pages give different figures (85% of the Fortune 100 on one, 80% on another).

Products as of October 2026

  • Avatar V. Launched on April 8, 2026, it builds a digital twin from a 15-second video. It is metered at 48 credits a minute and limited to three-minute videos inside Video Agent. More than 700 stock avatars are available on paid plans.
  • Video Translate. Translation of existing footage into 175+ languages and dialects, with Speed and Precision modes and batches of up to 100 jobs through the API. HeyGen's translation demos spread widely; one is covered in this video.
  • Video Agent. Aims to produce complete videos of up to 30 minutes from a prompt; plans allow videos up to 30 minutes on Creator and 60 on Business.
  • Bundled generative models. The Pro plan includes third-party video models such as ByteDance's Seedance 2.0 and Google's Veo 3.1 for B-roll.
  • Developer and agent access. A public API, a command-line tool, agent skills and a remote MCP server that works with Claude, Cursor, Gemini CLI and Manus (see MCP).
  • LiveAvatar. Real-time interactive avatars, sold as a separate product from $99 a month.

Paid plans start at $29 a month (Creator, or $24 billed yearly). Pro costs $49 a month ($41 yearly) with 4K output, and Business costs $149 a month plus $20 per seat, adding SSO, SCORM export and 60-minute videos.

HeyGen requires a consent video of under 30 seconds, reading a fixed statement and recorded at the same time as the avatar footage, before it creates a personal avatar. Its moderation policy bans depicting real people or public figures without explicit consent and restricts political content. It lists SOC 2 Type II, GDPR, CCPA and the EU-US Data Privacy Framework, supports SSO and SCIM, stores data in the US on AWS and excludes enterprise data from model training by default. As with any tool that can make a person appear to say anything, the risks covered in deepfakes apply, and published avatar video increasingly needs disclosure (see AI content labeling).

How It Compares

Against Synthesia, HeyGen offers more avatars, longer videos, a larger translation catalog, bundled generative models and a cheaper mid tier, and its API and MCP tooling are more developed. Synthesia is stronger on training workflows and certifications. Both are part of the shift toward digital humans as a production format, and both compete with voice platforms such as ElevenLabs on dubbing. See Synthesia vs HeyGen, AI dubbing and lip-sync and audio-to-video generation.

Further Reading