OmniHuman Models by ByteDance
ByteDanceProprietary
ByteDance releases · 23 in the last 12 months · this family litChangelog →
1 model2026
Last refreshed 2026-10-22. Next refresh: weekly.
Capabilities
VisionAll models
MultimodalAll models
Links
WebsiteAbout
ByteDance's audio-driven human/avatar video models (OmniHuman 1.0, 1.5, 2.0): one image plus audio in, lip-synced performance video out.
Decision facts
- Best fit
- video generationvision and multimodal work
- Capability starting point
- OmniHuman 2.0 with multimodal inputs
- Lowest tracked input
- Not tracked
- Closest related family
- BAGEL
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
1 in view
OmniHuman 2.0Current
Use when the workload needs video generation, multimodal inputs, and audio.
2026-10video generationmultimodal inputsaudio
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| OmniHuman 2.0 | Use when the workload needs video generation, multimodal inputs, and audio. | 2026-10 | video generationmultimodal inputsaudio | Current |
Release Timeline
1 release group2026-10
1 current
OmniHuman 2.0
Currentvideo generationmultimodal inputsaudio
Specifications(1 models)
| Model | Released | Vision | Multimodal |
|---|---|---|---|
| OmniHuman 2.0 | 2026-10 | Yes | Yes |
