Gemini Omni coming soonLearn more
EvoLink.AIEvoLink.AI

Wan 2.7 API

Generate and edit videos from text, images, references, and existing clips with the Wan 2.7 API — the next-gen open-source video model from Alibaba.
Model Type:
✓Text to Video (T2V)Image to Video (I2V)Reference Video (R2V)Video Edit (V2V)

Price: $0.086 - 0.144(~ 5.87 - 9.8 credits) per second of video

Highest stability with guaranteed 99.9% uptime. Recommended for production environments.

Use the same API endpoint for all versions. Only the model parameter differs.

0 (suggested: 2,000)
5s
2s15s

Click Generate to see preview

History

Max 20 items

0 running · 0 completed

Your generation history will appear here

Wan 2.7 API: Generate, Animate, Reference & Edit Video in One Endpoint

Wan 2.7 is the first Wan model that covers all four video workflows — text-to-video, image-to-video with first/last frame control, multi-character reference video with voice cloning, and instruction-based video editing — through a single EvoLink API call at $0.086/sec.

Wan 2.7 AI video generation and editing showcase

Pricing

WAN 2.7 Text to Video
Video Generation
Quality:720p
Price:
$0.088/ second
(6 Credits)
WAN 2.7 Text to Video
Video Generation
Quality:1080p
Price:
$0.147/ second
(10.02 Credits)

If it's down, we automatically use the next cheapest available—ensuring 99.9% uptime at the best possible price.

What You Can Build with Wan 2.7 API

Turn Scripts into Video Clips

Describe a scene — setting, subject, camera movement, mood — and Wan 2.7 generates a 720p or 1080p video up to 15 seconds. Add optional audio to sync video to music or voiceover. Use negative prompts to exclude unwanted elements. Ideal for ad creatives, social content pipelines, and automated video-first workflows.

Generate text-to-video
Wan 2.7 text to video generation showcase

Animate Product Images and Storyboards

Define both the first and last frame, and Wan 2.7 infers the motion trajectory between them — keeping subject identity stable with no drift. This is how teams turn product photos into scroll-stopping social clips, or convert storyboard frames into motion tests, without manual keyframing.

Animate images into video
Wan 2.7 image to video generation showcase

Build Multi-Character Video Series with Voice

Feed up to 5 reference images, videos, or audio clips into a single API call. Wan 2.7 locks each character's appearance and clones their voice from a 1-10 second sample. This means brand spokesperson videos, multi-character explainers, and episodic content series can maintain identity across clips without manual compositing.

Create reference video
Wan 2.7 reference video multi-character showcase

Edit Existing Videos Without Re-generating

Pass an existing clip and a text instruction — 'swap the background to a rain-soaked street', 'change the jacket to red', 'apply vintage film style' — and Wan 2.7 edits the video while preserving the original motion and structure. No other Wan version supports this. Iteration cycles that previously required full re-generation are now lightweight API calls.

Edit videos with AI
Wan 2.7 video editing showcase

Why Choose Wan 2.7 on EvoLink

Wan 2.7 is the only Wan model that unifies text-to-video, image-to-video, reference video, and video editing. EvoLink makes it cheaper and simpler to integrate.

$0.086/sec — Lower Than Direct API Providers

EvoLink routes Wan 2.7 at $0.086 per second of generated video, below the $0.10/sec charged by Together AI and other providers. A 10-second 720p clip costs under $0.86. 1080p runs at 1.67x the 720p rate. No subscriptions or minimum commitments.

Video Editing That No Other Wan Version Has

Wan 2.7 is the first Wan model with instruction-based video editing. Describe the change — swap a background, shift lighting, apply a style transfer — and the model edits the clip without re-generating from scratch. Wan 2.6 and 2.5 can only generate new videos, not edit existing ones.

One Integration for All Four Workflows

Text-to-video, image-to-video with first/last frame control, multi-character reference video with voice cloning, and video editing all go through /v1/videos/generations. Switch workflows by changing the model parameter. No separate SDKs, no separate billing — one EvoLink API key handles everything.

How to Use Wan 2.7 API Step by Step

Integrate Wan 2.7 into your app with a few API calls.

1

Get your API key

Sign up at EvoLink and generate an API key from the dashboard.

2

Choose your mode

Select wan2.7-text-to-video, wan2.7-image-to-video, wan2.7-reference-video, or wan2.7-video-edit as the model parameter.

3

Send your request

POST to /v1/videos/generations with your prompt, media inputs, and parameters. The API returns a task ID immediately.

4

Poll for results

GET /v1/tasks/{task_id} to check progress. When complete, download the video URL (valid for 24 hours).

View Wan 2.7 API docs

Wan 2.7 API Features

Everything you need for AI video generation and editing.

5000-Character Prompts

Write detailed scene descriptions with up to 5000 characters per prompt.

Negative Prompts

Exclude unwanted elements with negative prompts up to 500 characters.

First & Last Frame Control

Specify start and end frames for precise image-to-video animation.

Video Editing Mode

Edit existing videos with text prompts and up to 4 reference images.

Audio Integration

Input driving audio, reference voice, or keep original video sound.

720p & 1080p Output

Choose between standard and high-definition output quality.

Explore the Wan API family

Wan 2.7 is the latest flagship with text-to-video, image-to-video, reference video, and video editing. See how Wan 2.7 fits alongside Wan 2.6 for cinematic storytelling, Wan 2.5 for daily content volume, and Wan Image for text-to-image workflows.

View the Wan familyWan 2.6Wan 2.5Wan Image

Wan 2.7 API FAQs

Everything you need to know about the product and billing.

Wan 2.7 is the latest video generation model from Alibaba's Tongyi Wanxiang team. It supports four modes: text-to-video, image-to-video with frame control, multi-character reference video with voice cloning, and instruction-based video editing. EvoLink provides the Wan 2.7 API through a unified endpoint at $0.086 per second.
Wan 2.7 costs $0.086 per second of generated video at 720p, and 1.67x that rate ($0.144/sec) at 1080p. A 10-second 720p clip costs under $0.86. No subscriptions or minimum commitments — you pay only for what you generate.
wan2.7-text-to-video for text-to-video generation. wan2.7-image-to-video for image-to-video with first/last frame control. wan2.7-reference-video for multi-character reference video with voice cloning. wan2.7-video-edit for instruction-based editing of existing videos.
Wan 2.7 adds two capabilities that Wan 2.6 does not have: instruction-based video editing (wan2.7-video-edit) and multi-character reference video with voice cloning. Wan 2.7 also supports first-and-last-frame control in image-to-video mode, while Wan 2.6 supports first-frame only. Wan 2.6 remains a strong choice for cinematic multi-shot storytelling with Flash variants for faster iteration.
Yes. In reference video mode (wan2.7-reference-video), you can provide a 1-10 second audio clip, and the generated video's character speech will match the source speaker's vocal characteristics. Combined with up to 5 visual references, this enables multi-character scenes with consistent appearance and voice.
Send an existing video (2-10 seconds, mp4 or mov) via video_urls, describe the change in your prompt (e.g. 'change the background to a rain-soaked street'), and optionally provide up to 4 reference images for style guidance. The model edits the video without re-generating from scratch. Set duration to 0 to keep the original video length.
Yes. In image-to-video mode, use image_start for the first frame and image_end for the last frame. You can specify one or both. The model infers the motion trajectory between your two keyframes, keeping subject identity stable across the clip.
If you are using EvoLink, change the model parameter from wan2.6-text-to-video to wan2.7-text-to-video (or the corresponding variant). The API endpoint, authentication, and async task pattern remain the same. For reference video, wan2.7-reference-video adds voice cloning and multi-character support on top of what wan2.6-r2v provides.
Wan 2.7 uses a 27B parameter architecture with 14B active parameters via Mixture-of-Experts, released under Apache 2.0. Earlier versions like Wan 2.1 were also open-sourced. Check Alibaba's official announcements for the latest open-source status and weight availability.
Wan 2.7 supports 720p and 1080p output at 30fps. Video duration ranges from 2 to 15 seconds for text-to-video and image-to-video, 2 to 10 seconds for reference video, and 2 to 10 seconds for video editing. Prompts can be up to 5000 characters with 500-character negative prompts.
POST
/v1/videos/generations

Create Video

WAN 2.7 Text to Video (wan2.7-text-to-video) generates video from text prompts with optional audio input, negative prompts, and prompt enhancement.

Asynchronous processing mode, use the returned task ID to .

Generated video links are valid for 24 hours, please save them promptly.

Request Parameters

modelstringRequiredDefault: wan2.7-text-to-video

Video generation model name.

Examplewan2.7-text-to-video
promptstringRequired

Text description of the video to generate.

Notes
  • Maximum 5000 characters
ExampleA majestic eagle soaring through mountain peaks at sunset, cinematic lighting
negative_promptstringOptional

Describe what you do not want in the video.

Notes
  • Maximum 500 characters
Exampleblurry, low quality, distorted
audio_urlsarrayOptional

Audio URL array for video generation (driving audio). Only audio_urls[0] is used. The legacy single audio_url field is also accepted for backward compatibility.

Notes
  • Supported formats: MP3, WAV
  • Duration: 2-30 seconds
  • File size: max 15MB
  • If audio is longer than duration, it will be truncated
  • If audio is shorter, the remaining video will be silent
Example["https://example.com/audio.mp3"]
qualitystringOptionalDefault: 720p

Video output quality / resolution.

ValueDescription
720pStandard quality (1.0x price)
1080pHigh quality (1.67x price)
Example720p
aspect_ratiostringOptionalDefault: 16:9

Video aspect ratio.

ValueDescription
16:9Landscape video (default)
9:16Portrait video
1:1Square video
4:3Standard video
3:4Portrait standard
Example16:9
durationintegerOptionalDefault: 5

Video duration in seconds.

Notes
  • Range: 2-15 seconds
Example5
seedintegerOptional

Random seed for reproducible results.

Example42
prompt_extendbooleanOptionalDefault: false

Automatically enhance your prompt using AI for better results. Disabled by default on EvoLink to avoid silent prompt rewriting; pass true to opt in.

callback_urlstringOptional

HTTPS callback URL invoked when the task finishes (completed / failed / cancelled). Sent after billing confirmation.

Notes
  • HTTPS only. Internal IPs are rejected (127.0.0.1, 10.x.x.x, 172.16-31.x.x, 192.168.x.x).
  • Max URL length: 2048 chars.
  • Timeout: 10s; up to 3 retries with 1s/2s/4s backoff after failure.
  • A 2xx response is treated as success; other status codes trigger retry.
  • Callback body shape mirrors the task query endpoint response.
Examplehttps://your-domain.com/webhooks/video-task-completed

Request Example

curl -X POST "https://api.evolink.ai/v1/videos/generations" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "wan2.7-text-to-video",
  "prompt": "A majestic eagle soaring through mountain peaks at sunset",
  "quality": "720p",
  "aspect_ratio": "16:9",
  "duration": 5,
  "prompt_extend": false
}'

Response Example

{
  "created": 1757169743,
  "id": "task-unified-1757169743-abc123",
  "model": "wan2.7-text-to-video",
  "object": "video.generation.task",
  "progress": 0,
  "status": "pending",
  "task_info": {
    "can_cancel": true,
    "estimated_time": 60
  },
  "type": "video",
  "usage": {
    "billing_rule": "per_second",
    "credits_reserved": 10,
    "user_group": "default"
  }
}