# OmniHuman 2.0: $0.07/s Price, Specs, API & 5-Min Videos Source page: https://www.makeaivideo.net/omnihuman-2 > OmniHuman 2.0 turns one image and an audio track into a native 720p avatar video up to 5 minutes long. BytePlus API price $0.07/sec, down from $0.12. Plain-text version of this page for AI assistants and readers. Site overview: https://www.makeaivideo.net/llms.txt Prices and plan details change; the current ones are on https://www.makeaivideo.net/pricing --- [Home](https://www.makeaivideo.net/)/OmniHuman 2.0 # OmniHuman 2.0 — ByteDance's Audio-Driven Avatar Video Model One image and one audio track in, a native 720p video out where the character talks, sings and acts to the sound - up to 5 minutes in a single shot. Released October 15, 2026 · not available to generate here yet Demo videos: BytePlus (official OmniHuman 2.0 samples) ## Get an email if OmniHuman 2.0 opens here You can't generate with OmniHuman 2.0 on this site yet. Leave your email and we'll send one message if it opens, nothing else. ## What is OmniHuman 2.0? BytePlus describes OmniHuman 2.0 as a film-grade digital human model: it builds a video from a single image, an audio track and an optional text prompt. It comes from ByteDance and is available via BytePlus as an API. The character's lips, face, gestures and body follow the rhythm, delivery and meaning of the recording, and it handles several people in one scene. You supply the voice - it animates the image to match it. ## What's new in OmniHuman 2.0 ### 5 minutes in one shot Up to 5 minutes as a single continuous take, up from 60 seconds - no stitching clips together. ### Over 40% cheaper $0.07 per second, down from $0.12 for the previous tier. ### It understands the audio In BytePlus's words, it "doesn't just sync lips": it reads tone, emotion and the words in speech, music and dialogue. ### More motion and control Film-grade camera and body movement, multi-person scenes, and text prompts for actions and camera. Source: BytePlus on X, October 15, 2026 ## OmniHuman 2.0 specs and input limits - Developer: ByteDance, available via BytePlus - Released: October 15, 2026 - Inputs: One image + one audio track, optional text prompt - Output: Native 720p video - Length: Up to 5 minutes in one continuous shot (product page: 30s to 300s output) - Audio input: WAV or MP3, 4 to 300 seconds, under 15 MB (docs) - Image input: JPEG, PNG, WebP, BMP, TIFF, GIF, HEIC or HEIF, under 30 MB - Image size: Each side 300 to 6,000 px, aspect ratio 0.4 to 2.5 - Prompt languages: Chinese, English, Japanese, Korean, Spanish (Mexican), Indonesian - Prompt controls: Scene, movements and camera - API price: $0.07 per second Sources: BytePlus product page · OmniHuman 2.0 API docs ## OmniHuman 2.0 pricing BytePlus charges $0.07 per second of output. At that rate: 15 seconds $1.05 30 seconds $2.10 60 seconds $4.20 5 minutes $21.00 Rate × duration at the API list price as of October 15, 2026. Check BytePlus for current billing terms. ## OmniHuman 2.0 features ### Text prompt control Direct specific actions, camera moves and extra objects in the scene with a short prompt. ### Synchronized, lifelike people Lip movement, emotion and gestures stay in time with the audio, turning a still into a believable performance. ### Musical performances Digital singers with expressive motion and natural pauses, from quiet ballads to big stage songs - including high notes. ### Cinematic emotion Reads the context of the audio to drive gestures and shifts in emotion, so the delivery matches the words. ## OmniHuman 2.0 styles and use cases Demo videos: BytePlus (official OmniHuman 2.0 samples) ### Marketing and film E-commerce, advertising, brand marketing, short video and film & TV, with complex emotion and expressive dialogue. ### AI music videos Actions that follow the rhythm, emotion matched to the music, and dynamic camera movement. ### UGC and characters Cartoon and anime characters, pets, pixel-style visuals and virtual avatars - plus 3D-style characters. ### Fantasy vlogs Surreal scenes with realistic effects, believable motion, environmental interaction and controllable settings. ## OmniHuman 2.0 API overview - Asynchronous: submit a task, poll for the result, cancel if needed. - Endpoint at cv.byteplusapi.com, region ap-singapore-1, requests signed with HMAC-SHA256 using your BytePlus keys. - Inputs: an image URL, an audio URL and an optional prompt; the result is a video URL. - The result link is valid for 1 hour - save the file. Unfinished tasks expire after 12 hours. Full parameters in the OmniHuman 2.0 API docs . ## OmniHuman 2.0 vs the previous lip-sync tier Feature | OmniHuman 2.0 | Previous lip-sync tier | Price | $0.07 / second | $0.12 / second | Focus | Film-grade camera and body movement | Lip sync, avatar, vivid expression | Multi-person scenes | Yes | Not listed | Longest single shot | 5 minutes | 60 seconds | Both tiers are listed on the BytePlus product page; the 60-second figure is from BytePlus's launch post. ## Limits and safety - The API adds a watermark by default. - Inputs go through a content pre-check, and requests that fail it are rejected. - Use only your own likeness, or a person or character you're authorized to use. Don't make anyone appear to say or do something without their consent. ## OmniHuman 2.0 FAQ ### What is OmniHuman 2.0? ByteDance's audio-driven avatar video model, available via BytePlus. From one image, an audio track and an optional text prompt, it generates a film-grade video of that character performing the audio, with lip sync, expression, gestures and camera movement. ### When was OmniHuman 2.0 released? BytePlus announced it on October 15, 2026, as an API on its platform. ### How much does OmniHuman 2.0 cost? BytePlus lists $0.07 per second of video, down from $0.12 for the previous tier - over 40% cheaper. At that rate a 30-second clip works out to about $2.10 and a 5-minute clip about $21. ### How long can an OmniHuman 2.0 video be? Up to 5 minutes in a single continuous shot, up from 60 seconds, with no stitching. The API docs accept audio from 4 to 300 seconds; the product page describes 30 to 300 seconds of output. ### How is OmniHuman 2.0 different from the previous version? BytePlus says it doesn't just sync lips but understands the audio - tone, emotion and the words - for speech, music and dialogue. It also runs five times longer per shot (5 minutes vs 60 seconds) and costs over 40% less. ### How do I get API access? Through a BytePlus account. The API is asynchronous: you submit a task, poll for the result and can cancel it. Requests are signed with your BytePlus keys (HMAC-SHA256) in the ap-singapore-1 region. ### Which languages can the text prompt use? Chinese, English, Japanese, Korean, Spanish (Mexican) and Indonesian, per the API docs. The prompt is optional and steers the scene, movements and camera. ### Does it work with anime characters and pets? Yes. BytePlus shows cartoon and anime characters, pets, pixel-style visuals, 3D-style characters and virtual avatars among its examples. ### Is there an OmniHuman 2.0 benchmark or paper? Not yet. BytePlus hasn't published a 2.0 benchmark, and the technical report linked from the product page is the OmniHuman 1.5 one. ### Can I use OmniHuman 2.0 on this site? Not yet. Join the list on this page and we'll send one email if and when it opens here - nothing else. ### Whose photo can I use? Only your own likeness, or a person or character you're authorized to use. Making someone appear to say things they never said without their consent is not allowed. ## Try these in the meantime [Image to Video](https://www.makeaivideo.net/image-to-video) · [Text to Video](https://www.makeaivideo.net/text-to-video) · [All AI video models](https://www.makeaivideo.net/models) · [Trending templates](https://www.makeaivideo.net/templates)