The #1 text-to-video model

HeyGen Video is HeyGen's general-purpose AI video model, built on MiniMax H3 and post-trained by HeyGen. It generates a whole scene from a prompt, including subject, setting, light and sound, as a 5 to 15 second clip. It works in three modes: text to video, image to video and reference to video. On HeyGen, you can use it alongside your AI digital twin and other models in one workflow.

You can create brand film shots, product scenes, B-roll cutaways, social hooks and stylized explainers, each with matching audio. HeyGen Video holds scenes together: objects keep their size and position and obey physics. It follows a brief literally, so you get dialogue only when you ask for it. It performs best on short, contained shots with one subject, one place and one action.

HeyGen Video is available through the HeyGen API as the model heygen-video-1, on paid API keys. Create a key in your API settings, then send a prompt and a mode to the create endpoint. Finished clips appear in your workspace alongside your other HeyGen videos.

Pick a mode, write a detailed prompt and submit it. Use text to video for a new scene, image to video to open on your own first frame, or reference to video to keep a product, person or place you supply. Long prompts that name the camera, lens and sound beat short ones. Poll the job or set a callback, then download the MP4.

Clips run any length from 5 to 15 seconds at 480p, 768p, 1080p or 2K, encoded as 24 fps MP4 with stereo audio. Aspect ratios include 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16, so you can generate for YouTube, TikTok, Reels or Shorts without recropping. 1080p and 2K render in 16:9 or 9:16 only and bill at three times the 768p rate.

Yes. Videos generated with HeyGen Video can be used for commercial work, including ads, social posts, product launches and client deliverables, under your HeyGen commercial license. To keep invented branding out of frame, add "no logos, brand names, printed words or badges" to your prompt, since plain clothing and equipment can otherwise pick up made-up marks.

Yes. Room tone, sound effects and speech are generated with the picture from the same prompt, so the audio matches the space you described. Written dialogue fits at roughly 2.5 words per second, or about 37 words in a 15 second clip. Say "no music" when you do not want a music bed. For longer scripted delivery, your digital twin can speak in your own voice.

Seedance 2.0, and Veo 3.1 are known for complex, multi-shot cinematic motion. HeyGen Video competes on control: literal prompt following, sound from the same prompt, up to 12 references and repeatable seeds. The bigger advantage on HeyGen is workflow. You pick the right model for each shot, then add avatars, voice and captions in one place.

Yes. Set the seed yourself and the same prompt with the same seed, submitted in succession, returns an identical file. Hold the seed and change one clause at a time to iterate. When the seed is omitted the server picks a random one, so two identical requests return different videos. The completed job reports the seed it used.

Only if you let it. Prompt enhancement can reword the rest of the prompt, so set it to disabled when you want your text kept exactly as written. You can also pass a callback URL to be notified when a job finishes, so you do not have to poll for the result.