Start with text or a single image. Explore short-form video ideas, visual styles, and a straightforward 5–10 second workflow with Wan 2.5.
Example footage, not an account-generated result. Settings do not change this demo. No uploads or credits used.
Wan 2.5 is the baseline AI video model in the Wan series, designed for fast and efficient video generation with stable 1080p output. If you want to see output styles before generating, browse the Wan showcase.
Generate videos from text prompts or images with up to 1080p output. Supports a wide range of scenes, including single subjects, environments, stylized animation, and abstract visuals. Works with both English and Chinese prompts.
Reference footage, not a result generated in your account.
Wan 2.5 is optimized for speed, delivering faster generation times than higher-tier models. It’s ideal for testing ideas, iterating quickly, and scaling production workflows.
“Aerial view of a dense forest in autumn, slow tilt down from canopy to forest floor, golden afternoon light filtering through leaves, cinematic.”
Choose from multiple visual styles, including cinematic, realistic, anime, and abstract outputs. Easily match the visual tone to your creative needs.
Wan 2.5 supports natural language prompts in both English and Chinese, with strong performance in Chinese-language scenes and character generation.
Wan 2.5 is one of three currently accessible versions. Here is how it compares to the rest of the Wan series.
For current credit packs and generation costs across versions, see the pricing page.
| Feature | Wan 2.5 | Wan 2.6 | Wan 2.7 | Wan 3.0 |
|---|---|---|---|---|
| Max resolution | 1080p | 1080p | 1080p | 1080p |
| Text-to-Video | ✓ | ✓ | ✓ | ✓ |
| Image-to-Video | ✓ | ✓ | ✓ | ✓ |
| Motion consistency | Standard | Improved | Professional ✓ | Enhanced |
| Multi-shot support | — | — | ✓ | ✓ |
| Real-time preview | — | — | — | ✓ |
| Camera motion control | — | — | ✓ Advanced | Full (5 types) |
| Generation speed | Fastest | Fast | Standard | TBC |
| Audio generation | ✓ | ✓ | ✓ | ✓ (Improved) |
| API access | — | — | Not connected | Not connected |
| Status | UI preview | UI preview | UI preview | UI preview |
Wan 2.5 is the best choice when generation speed and lower resource cost matter more than resolution. For all other use cases, Wan 2.7 is the recommended version.
Wan 2.7 is the recommended version for most users — but Wan 2.5 has specific scenarios where it is the practical best choice. If you want to practice the broader workflow that carries into Wan 3.0, follow the step-by-step guide.
Wan 2.5's faster generation time means you can run more prompt iterations in the same time window. Use Wan 2.5 to find the right prompt direction, then switch to Wan 2.7 for the final 1080p output.
For workflows that require dozens of short clips per session — social content batches, dataset generation, background B-roll — Wan 2.5's lower resource cost per generation makes it the efficient choice.
For platforms, internal tools, and social media that require standard HD output. Wan 2.5 produces excellent visual quality at 1080p, making it an ideal choice for most current delivery standards.
Wan 2.5 is the lowest-friction entry point to the Wan platform. Its prompt conventions are identical to Wan 2.7 and Wan 3.0 — skills built here transfer directly to later versions. Start here, upgrade when you are ready.
Wan 2.5 uses the same prompt structure as all Wan versions. The formula is: Subject → Environment → Lighting → Camera → Style
Example 1 — Environment / Landscape
"Aerial view of a dense forest in autumn, slow tilt down from canopy to forest floor, golden afternoon light filtering through leaves, no subjects, cinematic"Example 2 — Character Scene
"A baker slides a tray of bread into a stone oven in a small French bakery, close-up tracking the movement, warm amber light, steam rising, realistic"Example 3 — Product
"A smartphone sitting on a dark minimal surface, slow orbit shot from above, single soft light from the left, commercial product aesthetic, clean"Example 4 — Anime Style
"A student looks out a rain-streaked classroom window at a grey sky, static shot slowly pushing in, melancholy mood, Makoto Shinkai color palette, anime""A woman in her 40s, silver-streaked hair, grey linen shirt" beats "a woman." Specificity drives consistency — especially important at 720p.
Wan 2.5 responds well to explicit camera direction: "slow push-in," "static wide shot," "overhead looking down," "handheld follow."
Because Wan 2.5 generates faster, you can test more prompt variations in less time. Treat each generation as a prompt experiment.
Wan 2.5 brings text-to-video and image-to-video workflows into one workspace, with controls for prompts, duration, and output resolution.
Released in early 2025, Wan 2.5 delivers text-to-video and image-to-video generation at up to 1080p resolution. You describe a scene in natural language or provide an image — subject, environment, lighting, camera movement, style — and the model renders it as a video. No stock footage, no traditional animation pipeline, no render farm. Just a prompt and a result.
Explore Wan 2.5, Wan 2.6, Wan 2.7, and Wan 3.0 through their dedicated workspace pages. The current demos let you compare workflows; live generation depends on the model service connected to this site.
The most common question on this page is whether to use Wan 2.5 or Wan 2.7. Here is the direct answer.
Use Wan 2.5 if:
Use Wan 2.7 if:
Focused: one primary subject, one clear environment, one camera movement. Specific: physical descriptions over abstract ones. Style-anchored: a single style reference significantly affects output consistency at HD resolutions.
WanAI offers one-time credit packs with no expiry. Review the current pricing before purchasing. Video generation availability and the credits required for each task will be shown in the workspace.
Wan 2.5 is an AI video model for text and image workflows with output options up to 1080p. This workspace lets you prepare prompts and preview examples; live video generation is not connected yet.
WanAI uses one-time credit packs. Purchased credits do not expire. Visit Pricing to compare the current packs; generation costs will be shown before you submit a video task.
Wan 2.5 supports up to 1080p (1920×1080) resolution. It also supports 720p and 480p for faster generation. Wan 3.0 also supports output up to 1080p, with longer 30-second workflows.
While both support 1080p and image-to-video, Wan 2.7 offers enhanced motion consistency, better prompt adherence, and multi-shot sequence support. Wan 2.7 is recommended for professional production work.
Yes. Wan 2.5 supports both text-to-video (T2V) and image-to-video (I2V), allowing you to animate existing images or generate new scenes from scratch.
No. Wan 2.5 is fully web-based. Open it in any modern browser, type your prompt, and generate. No app download, no installation, no local GPU required.
Yes. All Wan versions run on the same account. You can switch between versions at any time using the version selector in the generator. Credits are shared across all versions.
Stable · Live
Upgrade: 1080p + Image-to-Video added
Best for: Users who need 1080p but want a simple step up from 2.5
Recommended · Live
Upgrade: 1080p + I2V + Enhanced consistency + Multi-shot
Best for: Most creators — the best available version today
Now Live
Upgrade: 30-second video + Multi-subject consistency
Best for: Creators who need longer video workflows