Tasks and inputs
Wan-Animate-2 is listed for Image-to-video, Video-to-video. It accepts image, video, text and produces video. Use the publisher’s input format and preprocessing for your chosen deployment.
Wan-Animate-2 is Alibaba Wan's open 14B character-animation model that drives a character from video directly, with text-controlled camera angle.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Pricing conditions: No API pricing stated; open weights for self-hosting
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
What if the dancer in your reference clip was filmed from the front, but you want the animated character seen from the side? Wan-Animate-2 lets you say so in text. It separates the output camera viewpoint from the driving video, which earlier animation models could not.
Alibaba's Wan team released Wan-Animate-2 inference scripts and both Base and Distillation weights on 7 August 2026, under Apache 2.0.
Older character-animation systems first extract pose or motion with a separate model, then render. Wan-Animate-2 feeds the driving video straight into a redesigned diffusion transformer. Dropping the extractor removes a common source of jitter and identity loss, and lets the model use cues a skeleton never captured.
| Variant | Purpose |
|---|---|
| Wan-Animate-2 Base | Highest-fidelity motion and identity preservation |
| Wan-Animate-2 Distillation | Fewer steps for faster generation |
| Wan-Animate-2-Lite | Efficient variant aimed at real-time streaming animation |
The Hub repository Wan-AI/Wan2.2-Animate-2-14B and Diffusers versions (base and distilled) are available. Integrations exist for Diffusers, DiffSynth-Studio and ComfyUI. Example exports run at 24 fps.
The reference environment is Python 3.11 with PyTorch 2.7.0 on CUDA 12.6, plus FlashAttention. Diffusers support needed a source install at release, so check whether a tagged Diffusers version now includes it. The Hub repo is dated 14 July, but the weights were announced on 7 August; use the later date when you track versions.
It follows the earlier Wan2.2-Animate-14B, which is listed with the team's other releases on Wan-AI's Hugging Face listings. For prompt-led clips rather than motion transfer, Alibaba's hosted video models such as HappyHorse 1.1 R2V are the comparison.
Shoot three driving videos: front-on talking, a full-body turn, and fast footwork. Animate a stylised character and a photoreal one from each, then repeat with a different requested viewpoint. Score identity, hand quality and whether the new camera angle is physically plausible. Run the same set through the distilled weights to see what speed costs you. More options are in the image-to-video directory.
Compare the model on the work you need to complete.
Wan-Animate-2 is listed for Image-to-video, Video-to-video. It accepts image, video, text and produces video. Use the publisher’s input format and preprocessing for your chosen deployment.
Choose duration, resolution and frame rate together. Compare the cost of an accepted clip, including repeat generations.
The weights are downloadable under Apache-2.0. Check runtime support and size the complete model and working context for your expected concurrency.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
tbc
TBC mischief 1 is an image-to-video model built on Wan 2.2, sold on fal and OpenRouter at about half a cent per second for 3-5 s clips at up to 720p.
NVIDIA
Cosmos3-H-Dreams is NVIDIA's open real-time surgical world model: it streams Versius robot video from one endoscope frame and two-arm commands.
NVIDIA
Cosmos3-H-Embodied-Nano is NVIDIA's open 16B surgical world model that does forward dynamics, inverse dynamics and policy from one checkpoint.
NVIDIA
Cosmos3-Nano-Sim-Bimanual is NVIDIA's open 16B world model that simulates video of a two-armed AgiBot robot as it moves its camera and manipulates.
noiz-ai
Helix World (preview) is an open real-time world model from Noiz AI and HKUST that streams explorable 768x512 video with stereo sound that follows the camera.
Z.ai / Zhipu AI
SCAIL-2 is Zhipu's open character-animation model that drives a reference character from any video, with no pose skeletons or masks. MIT licence.