This guide has two parts: the Wan family of video models from Alibaba, and three tools that start from a video or a picture you already have. Other families: Veo and Gemini, Kling, and Seedance and Grok Imagine.
Wan 2.7
Four models, each for a different starting point, all priced per second.
- Wan 2.7 Text to Video makes video from text, with an optional audio URL, at 720p or 1080p, for 2 to 15 seconds.
- Wan 2.7 Image to Video animates from a first and last frame, supports clip continuation and driving audio, for 2 to 15 seconds.
- Wan 2.7 Reference to Video uses up to 5 images and videos as guidance for style and subject, for 2 to 10 seconds.
- Wan 2.7 Video Edit edits an existing video from a text prompt and an optional style reference, with 2 to 10 seconds of output.
Wan 2.6
- Wan 2.6 Video makes cinematic video from text, images or video with smooth motion and transitions, in clips of 5, 10 or 15 seconds, priced per clip.
- Wan 2.6 Flash Image to Video is the fast image to video model, with optional native audio, for 5, 10 or 15 seconds.
- Wan 2.6 Flash Video to Video is the fast video to video model, with optional native audio, for 5 or 10 seconds.
Tools for video you already have
- InfiniteTalk (MeiGen) makes a talking video from an image and audio, at 480p or 720p.
- Volcengine LipSync (ByteDance) is video to video lip sync with precise alignment to the voice.
- Topaz Video Upscale enhances video resolution up to 4×, keeping detail in every frame.
What each one costs
Prices are read from the SonnaLabs catalog.
| Model | Configuration | Price in credits |
|---|---|---|
 Wan 2.7 Text to VideoWan | Standard | 1,415 / second $0.099 |
| Wan 2.7 Image to VideoWan | Standard | 1,415 / second $0.099 |
| Wan 2.7 Reference to VideoWan | Standard | 1,415 / second $0.099 |
| Wan 2.7 Video EditWan | Standard | 1,415 / second $0.099 |
| Wan 2.6 VideoWan | 720p — 5s clip | 6,190 / generation $0.433 |
| 720p — 10s clip | 12,375 / generation $0.866 | |
| 720p — 15s clip | 18,565 / generation $1.30 | |
| 1080p — 5s clip | 9,280 / generation $0.650 | |
| 1080p — 10s clip | 18,565 / generation $1.30 | |
| 1080p — 15s clip | 27,845 / generation $1.95 | |
| Wan 2.6 Flash Image to VideoWan | Standard | 2,440 / generation $0.171 |
| Wan 2.6 Flash Video to VideoWan | Standard | 2,440 / generation $0.171 |
 InfiniteTalkMeiGen | Standard | 265 / second $0.019 |
 Volcengine LipSyncByteDance | Video-to-Video Lip Sync | 705 / second $0.049 |
 Topaz Video UpscaleTopaz Labs | Standard | 705 / second $0.049 |
Prices are in credits; the dollar amounts are estimates. The current prices are always on the models page.
How to choose
- Text only: Wan 2.7 Text to Video.
- You have a picture and want it to move: Wan 2.7 Image to Video, or Wan 2.6 Flash Image to Video for speed.
- You want the new clip to look like some pictures or clips: Wan 2.7 Reference to Video.
- A video that is almost right: Wan 2.7 Video Edit.
- Someone has to speak: InfiniteTalk starts from a picture and audio, and Volcengine LipSync fixes the mouth in a video you already have. For a presenter avatar, see Kling AI Avatar.
- A good clip that is too small: Topaz Video Upscale.
Video generation needs a paid plan or credit packs; see how credits work.