HeyGen Video API
HeyGen Video is HeyGen's generative engine that turns text prompts, images, or reference videos into complete scenes with visuals, dialogue, and sound effects.
HeyGen Video is a generative video model from HeyGen that builds complete scenes from scratch rather than assembling clips from a stock library. It accepts text prompts, images, and reference videos as input, and produces video with visuals, dialogue, and sound effects in a single pass. Output resolution ranges from 480p up to 2K, and the model supports fixed seeds plus up to 12 reference materials. It handles both text-to-video and image-to-video work, and generates sound effects natively.
The model responds best to shot-list style instructions: describe subject motion, camera setup, visual style, and audio or other constraints item by item to keep results controllable. That makes it well suited to product review videos such as unboxings, comparisons, and tutorials; advertising and marketing assets including social ads, brand films, and e-commerce product showcases; and education and training content, where avatar presenters can explain material in more than 175 languages. Teams can also stack shot descriptions to build storyboards for film-style scenes and short films, or apply it to vertical industries such as real estate and corporate communications.
HeyGen Video is API- and agent-friendly, with an MCP server, llms.txt, and typed schemas that let AI agents like Claude and Cursor call it directly. It also integrates into the Video Agent workflow, which automates scripting, scene construction, avatar selection, voiceover, and subtitles.
Updated:
MODEL ID
heygen-video-1- Output
- video
Highlights
- Ranks first on the Artificial Analysis video model leaderboard
- About 3.7s inference for a 10-second image-to-video clip
- Builds scenes from scratch with visuals, dialogue, and sound
- Handles both text-to-video and image-to-video
- Output from 480p up to 2K
- MCP server and typed schemas for AI agents
Providers
Compare API prices across providers. Select a provider to view all charges and conditions.
Benchmarks
- 第一
Artificial Analysis 视频模型排行榜
Scores as published by the model provider.
Video specifications
- Aspect ratios
- 21:9 / 16:9 / 4:3 / 1:1 / 3:4 / 9:16
- Audio · Output
- No
- Duration · Minimum
- 5 seconds
- Duration · Maximum
- 15 seconds
- Duration · Step
- 1 seconds
- Resolutions
- 480p / 768p / 2K
API reference
Model-specific protocol details are not published. Check the API reference for integration requirements.
API reference →Frequently asked questions
What is HeyGen Video?
HeyGen Video is a generative video model from HeyGen that builds complete scenes from scratch rather than assembling clips from a stock library. It accepts text prompts, images, and reference videos as input, and produces video with visuals, dialogue, and sound effects in a single pass. Output resolution ranges from 480p up to 2K, and the model supports fixed seeds plus up to 12 reference materials. It handles both text-to-video and image-to-video work, and generates sound effects natively. The model responds best to shot-list style instructions: describe subject motion, camera setup, visual style, and audio or other constraints item by item to keep results controllable. That makes it well suited to product review videos such as unboxings, comparisons, and tutorials; advertising and marketing assets including social ads, brand films, and e-commerce product showcases; and education and training content, where avatar presenters can explain material in more than 175 languages. Teams can also stack shot descriptions to build storyboards for film-style scenes and short films, or apply it to vertical industries such as real estate and corporate communications. HeyGen Video is API- and agent-friendly, with an MCP server, llms.txt, and typed schemas that let AI agents like Claude and Cursor call it directly. It also integrates into the Video Agent workflow, which automates scripting, scene construction, avatar selection, voiceover, and subtitles.
How is usage billed?
Your bill depends on the provider you use, its prices and your usage. Longer inputs or other conditions may have different prices. Billed based on the actual generation cost. See your bill for the final amount.
What inputs does HeyGen Video accept?
It accepts text prompts, images, and reference videos, and can use up to 12 reference materials in a single generation.
What does HeyGen Video output?
It outputs video that includes visuals, dialogue, and sound effects, at resolutions from 480p up to 2K.
What kinds of tasks is it suited for?
It fits product reviews, advertising and marketing assets, education and training content, film-style scenes and short films, and vertical applications such as real estate and corporate promotion.
How should prompts be structured?
Shot-list style instructions work best: describe subject motion, camera settings, visual style, and audio or constraints item by item for controllable results.
Can AI agents call HeyGen Video?
Yes. It provides an MCP server, llms.txt, and typed schemas so agents such as Claude and Cursor can invoke video generation directly.
How does it differ from other models in the HeyGen ecosystem?
HeyGen Video serves as the core generation engine, running alongside Seedance 2.0 and Veo 3.1 on the HeyGen platform, where users can pick a model per shot in the editor.