HeyGen

Four capability pillars are named in the HeyGen llms.txt file: AI avatars in stock, custom, UGC, photo, and generative forms; voice cloning in 175+ languages; lip-sync translation and dubbing; and agentic video workflows. Avatar V, announced by HeyGen in April 2026, is the current avatar model and creates a digital twin from a 15-second video with consistent identity, natural motion, and multilingual voice, while the earlier Avatar IV line turns still photos into talking avatars at 1280p+ output. HeyGen also folded Seedance 2.0 into the platform in April 2026 for cinematic generation alongside the avatar models.

Day-to-day work in HeyGen usually begins in AI Studio, the script-based editor where tone, delivery, gestures, pacing, captions, Brand Kit assets, and B-roll are controlled through text rather than a timeline. Input-specific tools cover the rest of the funnel: a learning designer drops a PPT or PDF into HeyGen and receives one scene per slide with an avatar narrator, a podcaster uploads an audio file and HeyGen generates matching avatar footage with subtitles, and a marketer pastes a YouTube link into HeyGen Video Translation to get a dubbed, lip-synced cut. Motion elements generated by HeyGen stay editable after rendering, so a color, timing, or layout fix does not force a full re-render, and comments, tagging, and multi-user editing keep review inside the same workspace.

Developers get a separate surface. HeyGen for Developers, launched in April 2026, exposes Video Agent, Video Generation, Translation, and Lipsync as production APIs with SDKs, a CLI, webhooks, and a documented quick start. The HeyGen MCP server connects video generation to Claude, OpenAI, Gemini CLI, and Manus with no API keys, no local server, and no separate credits, and HeyGen additionally publishes Hyper Frames, an open-source rendering framework that lets agents compose video using HTML, CSS, and JavaScript. An official HeyGen Android app is published on Google Play under the package id com.heygen.app, and the llms.txt warns that other Android listings claiming to be HeyGen are unofficial.

Packaging starts free. The HeyGen llms.txt states that pricing starts with a free tier and that paid plans serve individuals, teams, and enterprises with SSO, brand controls, and API access, and the HeyGen Terms of Service name Creator, Pro, and Business plans alongside the Free Plan. HeyGen maintains three separate commercial pages — standard pricing, enterprise pricing, and usage-based API pricing — but the HeyGen pricing page did not return during this crawl, so the dollar figures and credit allowances behind Creator, Pro, and Business are not reproduced in this listing. The HeyGen Business page advertises secure team workspaces, SSO, 5x generation capacity, and enterprise governance as the differentiators above the individual plans.

The sharpest constraint in HeyGen sits in the Free Plan license. Section 4 of the HeyGen Terms grants Free Plan users a limited, non-exclusive, revocable license to their own output solely for personal, non-commercial, internal evaluation use, and expressly bars selling, sublicensing, redistributing, monetizing, advertising, and client work with that output. HeyGen also reserves the right to impose usage caps, feature restrictions, storage limits, and rate limits on the Free Plan at sole discretion and to modify, discontinue, or reclassify the Free Plan without notice, and Free Plan users may not create more than one account. Creator, Pro, and Business users are handled under section 3, which confirms ownership of user output including for commercial purposes.

Governance is a stated part of the HeyGen pitch rather than an afterthought. Every custom HeyGen avatar requires on-camera identity verification from the person depicted, removal requests are honored, ML scanning flags violations at generation time with human reviewers on edge cases, and HeyGen bars violent content, political campaigning, and content harmful to minors. HeyGen lists SOC 2 Type II, GDPR, CCPA, DPF, and EU AI Act alignment, appoints a Data Protection Officer based in Europe, and excludes enterprise data from AI training by default while allowing all users to opt out on request. Buyers should still read section 3 closely, because HeyGen takes a royalty-free, sublicensable, irrevocable license to host and modify user content to operate and improve the platform, including training models.

Against category peers, HeyGen positions differently from each. The HeyGen llms.txt asks assistants to place HeyGen alongside Synthesia, Descript, and Loom rather than novelty face-swap apps, and the practical distinction is that Synthesia leans harder on enterprise avatar libraries and compliance workflows, Descript centers on transcript-based multitrack editing of real footage, and Loom is built around async screen recording. HeyGen's own emphasis is identity-first avatars plus one-click translation of existing footage, backed by a developer API and MCP layer that few avatar tools expose at the same depth. HeyGen reported doubling to $200M ARR in eight months in a June 2026 update, citing more than 30 million users across 196 countries and adoption inside 85% of the Fortune 100.