VIBE Blog8 min read

WAN 2.2 AI Video Generator: Cheap 720p Draft Clips

WAN 2.2 turns a prompt or photo into a 5-second 720p clip for just 15 to 20 tokens. See its exact limits in VIBE, how it compares with WAN 2.6 and 2.7, and when to use it.

Jiyeon Kim

Jiyeon Kim

AI Video Editor at VIBE

WAN 2.2 AI video generator creating a cheap 720p draft clip from a text prompt
On this page
  1. What Is the WAN 2.2 AI Video Generator?
  2. What Can You Make With WAN 2.2 in VIBE?
  3. How Do I Make an AI Video With WAN 2.2?
  4. Can I Turn a Photo Into a Video With WAN 2.2?
  5. How Much Does WAN 2.2 Cost in VIBE?
  6. What Is the Difference Between WAN 2.2, WAN 2.6 and WAN 2.7?
  7. Which AI Video Model Is Best for Cheap Test Clips?
  8. How Do I Write Better Prompts for WAN 2.2?
  9. Frequently Asked Questions
  10. Conclusion

The WAN 2.2 AI video generator in VIBE is the fast, low-cost way to turn a prompt or a photo into a 5-second 720p clip. WAN 2.2 is an Alibaba model that costs 3 tokens per second from text and 4 tokens per second from an image, so a full clip is 15 or 20 tokens. VIBE is an AI video generator app that lets you create stunning videos from text prompts or images using the latest AI models like Kling, Sora, and Veo.

WAN 2.2 is not the newest WAN model, and that is exactly why it is useful. It is the draft model: you test an idea cheaply, then move the winning prompt to a bigger model. This guide covers what WAN 2.2 does, its exact limits in VIBE, what it costs, and when to pick WAN 2.6 or WAN 2.7 instead.

WAN 2.2 AI video generator turning a text prompt into a 720p clip
WAN 2.2 AI video generator turning a text prompt into a 720p clip

What Is the WAN 2.2 AI Video Generator?

WAN 2.2 is an open video generation model from Alibaba, released in late July 2025 under the Apache 2.0 license. Alibaba describes it as its first video model built on a Mixture-of-Experts design, where one expert handles the early, noisy stages of generation and a second expert refines detail. Alibaba also says it expanded the training data by 65.6% more images and 83.2% more videos compared with WAN 2.1.

Inside VIBE, WAN 2.2 runs the compact 5B variant, tuned for speed. That is why it is quick and cheap, and also why it is not the sharpest or longest model in the app.

Here is what WAN 2.2 offers in VIBE:

  • Text-to-video and image-to-video in one model.
  • 720p output at 16:9 or 9:16.
  • Fixed 5-second clips.
  • No audio track, so you add sound separately.
  • 3 tokens per second from text, 4 tokens per second from an image.

For the broader history of the family, see the WAN AI video generator guide.

What Can You Make With WAN 2.2 in VIBE?

WAN 2.2 is best at short, simple scenes with one clear subject and one clear motion. Think a slow camera push toward a character, steam rising from a cup, waves rolling onto a beach, or a product turning on a table.

It handles these well because five seconds is enough for a single action. It struggles when you ask for crowds, fast sports, readable text, or three actions in a row. Those are better suited to a larger model.

Good fits for WAN 2.2:

  • Concept tests before spending tokens on a premium model.
  • Looping background clips for social posts.
  • Animated stills from a photo.
  • Quick style experiments, such as watercolor, clay, or neon night scenes.
A creator reviewing several short WAN 2.2 draft clips on a monitor
A creator reviewing several short WAN 2.2 draft clips on a monitor

How Do I Make an AI Video With WAN 2.2?

You make a WAN 2.2 video by choosing the model, writing one short prompt, picking an aspect ratio and generating. VIBE runs in your browser on any phone or computer, so there is nothing to install.

  1. Sign in at app.vibeclips.ai and open the text-to-video screen.
  2. Select WAN 2.2 in the model list.
  3. Write a prompt with a subject, an action and a camera move.
  4. Choose 16:9 for YouTube and desktop, or 9:16 for Shorts, Reels and TikTok.
  5. Generate, review the 5-second clip, and download it when you like it.

A reliable prompt formula is subject, action, setting, light, camera. For example: "A red fox walks through fresh snow at dawn, soft golden light, slow dolly-in, shallow depth of field."

Make your first AI video in 60 seconds

Generate AI videos with Kling, Veo, Sora and more, right in your browser.

Can I Turn a Photo Into a Video With WAN 2.2?

Yes. WAN 2.2 supports image-to-video, so you can upload a photo and describe how it should move. It costs 4 tokens per second, which is 20 tokens for a full 5-second clip.

Image-to-video works best when the photo already has a clear subject and some space around it. Describe only the motion, not the picture, because the model can already see the picture. "The camera slowly pushes in while hair moves in the wind" is better than a long description of the person.

If you want more control over a single photo, the image-to-video tools and the photo-to-video guide show how to prepare images.

How Much Does WAN 2.2 Cost in VIBE?

A WAN 2.2 clip costs 15 tokens from text and 20 tokens from an image, which makes it one of the cheapest models in the catalog per clip. VIBE has no free plan; you buy a one-time token pack or choose a monthly or yearly plan that unlocks every model.

Here is how WAN 2.2 compares with the newer WAN models in token cost:

  • WAN 2.2: 3 tokens per second, so 15 tokens for 5 seconds.
  • WAN 2.6: 10 tokens per second at 720p, so 50 tokens for 5 seconds, or 15 tokens per second at 1080p.
  • WAN 2.7: 10 tokens per second, from 2 up to 15 seconds.

That means one WAN 2.6 clip costs more than three WAN 2.2 drafts. For the full pricing picture, read how AI video generator app pricing works.

Token cost comparison of WAN 2.2, WAN 2.6 and WAN 2.7 shown as glowing bars
Token cost comparison of WAN 2.2, WAN 2.6 and WAN 2.7 shown as glowing bars

What Is the Difference Between WAN 2.2, WAN 2.6 and WAN 2.7?

WAN 2.2 is the fast 5-second 720p draft model, WAN 2.6 adds 1080p, longer clips and built-in audio, and WAN 2.7 adds flexible lengths and audio sync. The differences come straight from the VIBE model list.

  • WAN 2.2: 720p, 5 seconds, 16:9 and 9:16, no audio.
  • WAN 2.6: 720p or 1080p, 5, 10 or 15 seconds, 16:9 and 9:16, audio always on, and multi-shot scene transitions for longer storytelling.
  • WAN 2.7: 720p, any whole second from 2 to 15, five aspect ratios including 1:1, 4:3 and 3:4, audio always on, and the option to upload a voice or music track to sync the video to it.

Choose WAN 2.2 when cost and speed matter most. Choose WAN 2.6 when you need 1080p or sound. Choose WAN 2.7 when you need a specific length or a square or portrait frame.

Which AI Video Model Is Best for Cheap Test Clips?

For cheap test clips, WAN 2.2 is a strong first stop because 15 tokens buys a complete 5-second 720p preview. The best workflow is a draft-then-upgrade loop.

  1. Generate three to five variations with WAN 2.2 using slightly different prompts.
  2. Keep the one with the best motion and framing.
  3. Re-run that exact prompt on WAN 2.6, WAN 2.7, Kling or Veo for the final render.

This is the same logic a film crew uses with storyboards. The cheap pass finds the idea, and the expensive pass delivers the polish. To compare the larger models, see how AI video generation models differ and the models section at /#models.

A storyboard of draft clips upgrading into a polished final video
A storyboard of draft clips upgrading into a polished final video

How Do I Write Better Prompts for WAN 2.2?

Write one subject, one action and one camera move, because a 5-second clip cannot hold more. Short prompts of 20 to 40 words usually beat long ones.

Tips that work well:

  • Name the camera move: slow dolly-in, orbit, handheld, static wide shot.
  • Describe light and time of day: golden hour, neon night, overcast morning.
  • Keep motion gentle for the cleanest results: drifting, swaying, flowing.
  • Avoid text on signs and screens, because small writing often warps.
  • Use 9:16 only when you plan to post vertically.

Since WAN 2.2 has no audio, plan the sound separately. If you need speech or music inside the clip, WAN 2.6 and WAN 2.7 generate audio automatically.

Frequently Asked Questions

What is VIBE?

VIBE is an AI video generator app that lets you create stunning videos from text prompts or images using the latest AI models like Kling, Sora, and Veo. It runs in your browser, so you can use it on any phone or computer.

Is WAN 2.2 available in an AI video generator online?

Yes. WAN 2.2 is available inside VIBE in your browser. You select it from the model list and generate 5-second 720p clips from text or from a photo.

How long can a WAN 2.2 video be?

A WAN 2.2 clip in VIBE is always 5 seconds long. For 10 or 15 seconds, use WAN 2.6, and for any length from 2 to 15 seconds, use WAN 2.7.

Does WAN 2.2 generate audio?

No. WAN 2.2 produces silent video in VIBE. WAN 2.6 and WAN 2.7 generate audio as part of the clip.

How many tokens does a WAN 2.2 video cost?

It costs 3 tokens per second from text and 4 tokens per second from an image. A 5-second clip is therefore 15 or 20 tokens.

Can I use WAN 2.2 for TikTok or Instagram Reels?

Yes. Choose the 9:16 aspect ratio for vertical clips, then download the finished video and post it. The clip is 720p, which suits mobile feeds.

Is WAN 2.2 better than WAN 2.6?

No, it is cheaper and faster, not better. WAN 2.6 offers 1080p, longer clips and audio, so it is the stronger final-render model.

Conclusion

The WAN 2.2 AI video generator is the budget draft model in VIBE: 5 seconds, 720p, text or image input, and only 15 to 20 tokens per clip. Use it to test prompts quickly, then move the best idea to WAN 2.6, WAN 2.7 or another model for the final version.

VIBE is an AI video generator app that runs in your browser and puts multiple AI models in one place. Open app.vibeclips.ai, pick WAN 2.2, and generate your first draft today.

Share this article

View all articles →

Make your first AI video in 60 seconds

Generate AI videos with Kling, Veo, Sora and more, right in your browser.