Wan 2.5 on Novita AI: The Future of AI Video Creation

Wan 2.5, the latest proprietary evolution of the Wan video generation series, is now officially available on Novita AI. This update introduces two dedicated models:

• Wan 2.5 Preview Text-to-Video (T2V) — Generates high-quality videos directly from text descriptions, supporting 5-second and 10-second durations.
• Wan 2.5 Preview Image-to-Video (I2V) — Creates smooth motion videos from an initial image and text guidance, also available in 5-second and 10-second formats.

Both variants debut new audio capabilities, offering automatic dubbing or the option to upload a custom audio file for seamless synchronization. Built on advanced deep learning, Wan 2.5 delivers sharper visuals, smoother motion, and faster generation—empowering creators, educators, and marketers to produce professional-grade video content with minimal effort.

Here’s the current Wan 2.5 pricing on Novita AI:

• Text-to-video mode: 5-second videos at 1080P cost $0.75 each
• Image-to-video mode: 5-second videos at 1080P cost $0.75 each

Key Highlights of Wan 2.5

Wan 2.5 builds upon the solid foundation of Wan 2.1, delivering significant enhancements in both generation quality and overall model capability. These advancements stem from several core technical innovations, most notably the integration of a Mixture-of-Experts (MoE) architecture, the expansion and refinement of training data, and the adoption of high-compression video generation techniques.

1. AI-Driven Automation
Automatically handles scene transitions, lighting, and timing—minimizing manual editing while ensuring cinematic consistency.

2. Text-to-Video & Image-to-Video Generation
Transforms prompts or scripts into dynamic, ready-to-share videos, drastically cutting production time.

3. Precise Audio Synchronization
Aligns voiceovers, music, and sound effects perfectly for professional-grade audiovisual results.

4. High-Definition Output
Supports export up to 1080p and 4K, delivering sharp, vibrant visuals suitable for all platforms.

5. Flexible Templates & Customization
Offers an expanded library of templates with adjustable effects and branding options for maximum creative freedom.

6. Hardware-Friendly Efficiency
Runs smoothly even on standard PCs and consumer GPUs—ideal for individual creators or small teams.

Wan 2.5 integrates the Mixture-of-Experts (MoE) architecture into its video diffusion framework. Proven effective in large language models, MoE enables a substantial increase in total model parameters without significantly raising inference costs. In Wan 2.5’s A14B series, a dual-expert configuration is introduced specifically for the diffusion denoising process: a high-noise expert manages early-stage generation to establish overall composition, while a low-noise expert handles later stages to enhance fine details. Each expert contains approximately 14B parameters, bringing the total parameter count to 27B with only 14B active at any given step,maintaining nearly the same inference workload and GPU memory usage.

Wan 2.5 vs. Wan 2.2: What’s new?

While both belong to the same model family, Wan 2.5 introduces major upgrades that enhance performance, efficiency, and user experience across the board. It builds upon Wan 2.1’s foundation with faster rendering, improved realism, and expanded creative control.

Comparison	Wan 2.5	Wan 2.2
Algorithm	Employs advanced deep learning for higher accuracy and lifelike visuals	Utilizes standard deep learning, moderate video quality
Performance	Offers faster generation speed and efficiently handles complex, high-resolution scenes	Generates videos slowly and supports limited high-res output
Features	Adds expanded templates, precise audio sync, and customizable effects	Provides only basic templates and effects
User Experience	Intuitive, user-friendly design suitable for both beginners and professionals	Conventional interface requiring some learning

Applications of the Wan 2.5 AI Video Model

• Marketing Videos: Effortlessly produce professional promotional content for products, campaigns, or brand storytelling.
• Social Media Content: Generate captivating videos tailored for platforms like Instagram, TikTok, and YouTube to boost engagement.
• Educational Resources: Transform lesson plans, tutorials, or lectures into interactive and visually compelling learning materials.
• Corporate Presentations: Create polished internal communications and training videos quickly and efficiently.
• Creative Projects: Perfect for animations, narrative-driven content, and multimedia art, enabling limitless creative expression.

Wan 2.5 Video Generation Showcase

T2V Prompt: A young man sits still on a subway train, surrounded by blurred figures moving rapidly. [Close-up] His eyes, barely blinking, intensify the sense of loneliness.

I2V Image:

I2V Prompt: A queen with silver hair stands solemnly in a medieval fortress courtyard. Her cloak flows gently in the wind as soldiers stand silently behind her. The camera slowly zooms in, capturing the subtle motion of her hair and the flickering light on her armor. A cinematic tone with soft morning light and dramatic atmosphere.

Get Started with Wan 2.5 on Novita AI!

Wan 2.5 is now available on Novita AI at just $0.75 per 5-second 1080p video (both T2V and I2V), so you can explore powerful video generation while keeping costs predictable.

Step 1: Log In and Access the Model Library

Log in to your account and click on the Model Library button.

Step 2: Choose Your Model

Browse through the available options and select the model that suits your needs.

Step 3: Get Your API Key

To authenticate with the API, we will provide you with a new API key. Entering the setting page, you can copy the API key as indicated in the image.

Step 4: Install the API

Install API using the package manager specific to your programming language.

After installation, import the necessary libraries into your development environment. Initialize the API with your API key to start interacting with Novita AI LLM. This is an example of using chat completions API for python users. Check the Docs for more details.

Wan 2.5 Image to Video API Example

import requests

url = "https://api.novita.ai/v3/async/wan-2.5-i2v-preview"

payload = {
    "input": {
        "prompt": "<string>",
        "negative_prompt": "<string>",
        "img_url": "<string>",
        "audio_url": "<string>"
    },
    "parameters": {
        "resolution": "<string>",
        "duration": 123,
        "prompt_extend": True,
        "audio": True,
        "seed": 123
    }
}
headers = {
    "Content-Type": "<content-type>",
    "Authorization": "<authorization>"
}

response = requests.post(url, json=payload, headers=headers)

print(response.json())

By combining intelligent automation with flexible creative control, Wan 2.5 redefines what’s possible in AI-powered video generation. Its dual Text-to-Video and Image-to-Video models, enhanced by precise audio integration and accelerated rendering, make professional video creation faster, smoother, and more accessible than ever. Whether used for storytelling, education, or marketing, Wan 2.5 empowers every creator to bring ideas to life effortlessly—all within a single, streamlined platform.

Frequently Asked Questions

Novita AI is the All-in-one cloud platform that empowers your AI ambitions. Integrated APIs, serverless, GPU Instance — the cost-effective tools you need. Eliminate infrastructure, start free, and make your AI vision a reality.