Wan 2.2 T2V A14B
Model Information
Display Name: Wan 2.2 T2V A14B
API Model ID: wan-ai/wan2.2-t2v-a14b
Category: Text To Video
Description: Wan 2.2 T2V A14B is Wan-AI's open-source text-to-video model with 27 billion total parameters using a novel two-expert Mixture-of-Experts architecture, activating 14 billion parameters per step for efficient high-quality video generation. **Key Features:** - 480P and 720P video generation - 2 or 5 second video clips at 16fps - Novel two-stage denoising (high-noise + low-noise experts) - Apache 2.0 license **Capabilities:** - Text-to-video generation - Negative prompt support - Seed control for reproducible outputs - Multi-resolution support **Best For:** - Short video clip generation from text - Creative content production - Rapid video prototyping - Marketing and social media video content **Technical Specs:** - Architecture: Mixture-of-Experts (2 experts) - Total Parameters: 27 billion - Active Parameters: 14 billion per step - High-noise expert: early denoising stages - Low-noise expert: later refinement stages - Frame Rate: 16fps - License: Apache 2.0 **Performance:** - Superior results on Wan-Bench 2.0 evaluation metrics - Improved visual coherence and detail - Supports multi-GPU inference (FSDP + DeepSpeed Ulysses)
How to Use This Model
To use Wan 2.2 T2V A14B via the HInow.ai API, use the model ID: wan-ai/wan2.2-t2v-a14b
API Request Example (Chat/Text)
POST https://api.hinow.ai/v1/chat/completions
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json
{
"model": "wan-ai/wan2.2-t2v-a14b",
"messages": [
{"role": "user", "content": "Your message here"}
]
}
Pricing
- total: $0.054
Available Parameters
- resolution: Output video resolution (default: 480p) (Options: 480p, 720p)
- duration: Video duration in seconds (default: 5) (Options: 2, 5)
- negative_prompt: Text describing what to avoid in the video
Quick Reference
To use this model, set: "model": "wan-ai/wan2.2-t2v-a14b"
Featured: Yes
Documentation: https://hinow.ai/models/wan-ai/wan2.2-t2v-a14b
API Endpoint: https://api.hinow.ai/v1
