# Wan v2.6 Reference-to-Video Flash

Wan v2.6 Reference-to-Video Flash is the speed-optimized version of Alibaba Cloud's reference-to-video model, delivering character-consistent video generation from reference clips in a fraction of the time required by the standard R2V model.

- **Model ID:** `alibaba/wan-v2.6-r2v-flash`
- **Type:** video
- **Providers:** alibaba
- **Context window:** Not applicable
- **Maximum output tokens:** 0
- **Pricing:** See the provider details below
- **Canonical page:** https://vercel.com/ai-gateway/models/wan-v2.6-r2v-flash

## Supported parameters

- **Parameters:** `aspectRatio`, `resolution`, `duration`, `fps`
- **Operations:** `reference-to-video`
- **Input modalities:** `text`, `image`, `video`
- **Output modalities:** `video`
- **Aspect ratios:** `16:9`, `9:16`, `1:1`, `4:3`, `3:4`
- **Resolutions:** `720p`, `1080p`
- **Durations:** `5s`, `10s`
- **Frame rates:** `24 fps`
- **Maximum outputs per request:** Not reported
- **Maximum input images:** 5
- **Input image formats:** `jpeg`, `jpg`, `png`, `bmp`, `webp`
- **Maximum input image size:** 20 MB

## Example

```ts
import { experimental_generateVideo as generateVideo } from 'ai';

const result = await generateVideo({
  model: 'alibaba/wan-v2.6-r2v-flash',
  prompt: 'A serene mountain lake at sunrise.'
});
```

## About

Wan v2.6 Reference-to-Video Flash shares the same core capability as the standard R2V model, extracting a subject's visual and vocal identity from a reference video and placing them into text-prompted new scenes, but is built for contexts where generation speed is the binding constraint. The Flash architecture reduces wait times substantially, enabling creative teams to iterate through prompt variations and scene configurations much faster than the standard model allows.

The reference extraction pipeline works the same way: pass reference URLs alongside a descriptive prompt, and use `character1`, `character2`, and so on in the instruction to match URL order (images or videos; 2 to 30 seconds per video reference, within provider limits on total references). The model extracts appearance, movement style, and voice characteristics and applies them to the generated output. Because the Flash variant is optimized for speed rather than peak reconstruction fidelity, fine details in identity preservation may differ slightly from the standard R2V output.

R2V Flash supports 720p and 1080p resolutions and output durations of 2 to 10 seconds, making it suitable for social content previews, storyboard animatics, or any workflow where a director needs rapid visual confirmation that a scene concept works before ordering full-quality renders.

## What to consider

When a production deliverable demands maximum identity fidelity from the reference material, evaluate the standard wan-v2.6-r2v model before finalizing your pipeline choice.

## When to use

### Best For

- Fast iteration on scene concepts where character identity needs to be verified before committing to full-quality R2V renders
- Social media content pipelines where volume and speed of character-consistent clips outweigh pixel-perfect fidelity
- Storyboard and animatic generation for live-action or animation pre-production using real talent references
- High-throughput brand content creation where a mascot or spokesperson must appear across many generated assets quickly

### Consider Alternatives When

- If final delivery quality and maximum identity fidelity are required, use wan-v2.6-r2v for the highest-quality character transfer
- If your source material is a still photo rather than a video clip, use wan-v2.6-i2v-flash for image-based animation
- If no reference subject is needed, use wan-v2.6-t2v for purely text-prompted video generation

## Best for

- **Fast scene concept iteration:** Verifying character identity before committing to full-quality R2V renders
- **High-volume social content:** Pipelines where volume and speed of character-consistent clips outweigh pixel-perfect fidelity
- **Storyboards and animatics:** Live-action or animation pre-production using real talent references
- **High-throughput brand content:** Generating many assets quickly where a mascot or spokesperson must appear consistently

## Consider alternatives

- **Maximum identity fidelity:** Use wan-v2.6-r2v for the highest-quality character transfer in final deliveries
- **Still-photo source material:** Use wan-v2.6-i2v-flash for image-based animation when the source is a still photo
- **No reference subject needed:** Use wan-v2.6-t2v for purely text-prompted video generation

## Frequently asked questions

### What makes R2V Flash different from the standard R2V model?

Flash is speed-optimized. It generates reference-consistent video much faster than the standard R2V at a potential tradeoff in peak identity fidelity. For drafts and iteration, Flash is preferred; for final output, the standard R2V model is recommended.

### Does R2V Flash support the same reference video format as standard R2V?

Yes. Both accept the same reference URL lists and prompt conventions: use `character1`, `character2`, and so on in the prompt, in URL order, with 2 to 30 seconds per video reference where applicable.

### What resolutions does R2V Flash support?

720p and 1080p. The R2V variants don't include a 480p option.

### What is the maximum generated video length?

Output duration is 2 to 10 seconds for Wan R2V on AI Gateway. The 15-second option available on some T2V and I2V models does not apply here.

### Can R2V Flash handle multiple characters from different reference clips in one scene?

Yes. You can combine several reference URLs in one request (within provider limits) and name them `character1`, `character2`, and so on in the prompt.

### Is audio included in generated output?

Voice and audio characteristics captured from the reference clips are part of the identity extraction process; check provider-level documentation for specific audio output behavior.

## Links

- [Model page](https://vercel.com/ai-gateway/models/wan-v2.6-r2v-flash)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://wan.video/)
- [Provider pricing](https://modelstudio.console.alibabacloud.com/ap-southeast-1/?tab=doc#/doc/?type=model&url=prices)