Video Generation API is now live!
Wan 2.7Video API

Wan 2.7 APIfor developers.

Generate cinematic text-to-video, first-and-last-frame image-to-video, multi-reference scenes, and prompt-driven video edits with Alibaba's Wan 2.7 stack.Choose direct fal.ai endpoints, WaveSpeed REST models, or OpenRouter's routed video API based on how much control or abstraction you want.

see more

Creative Direction

Visual references for commercial API workflows

These stills come from external IMA creative assets and are used here as art direction reference for image-led or campaign-style motion workflows.

Humpback whale reference image used in the fal.ai Wan 2.7 image-to-video example.

fal.ai Example Input

Whale First Frame

fal.ai's Wan 2.7 image-to-video docs use this whale frame as the input still. It is a good example of the kind of clean, well-lit source image that helps frame-guided motion look deliberate instead of noisy.

Cinematic atmosphere still used as a visual reference for Wan 2.7 prompting.

Scene Direction

Atmosphere Reference

Use a cinematic still like this when you want Wan 2.7 prompts to lock onto camera mood, texture, and lighting style before you move into text-to-video or reference-driven generation.

Modes

T2V, I2V, R2V, video edit

One model family for prompt-only generation, frame-guided motion, reference-driven scenes, and prompt-led video edits

Resolution

720p and 1080p

All cited provider pages expose 720p and 1080p output tiers for current Wan 2.7 video workflows

Duration

2 to 15 seconds

Text-to-video and image-to-video go up to 15 seconds, while reference-to-video and video edit are usually capped earlier

Reference control

Single frame to multi-ref

Start from one image, set first and last frame, or combine multiple videos and images for identity-consistent scenes

Audio

Optional input or routed support

fal.ai text-to-video and image-to-video accept audio input, WaveSpeed edit exposes audio handling, and OpenRouter normalizes audio generation flags

Pricing

From $0.10 / second

As of April 2026, WaveSpeed and OpenRouter list $0.10 / second for common 720p Wan 2.7 video runs, while reference-to-video carries higher reference-processing cost

Available Endpoints

Start building with the Wan 2.7 API

Multiple endpoints for text-to-video, image-to-video, fast preview flows, and async job retrieval. This section is laid out more like a product catalog than raw docs so users can scan what to use first.

NewDirect

Endpoint

fal.ai Text-to-Video

fal-ai/wan/v2.7/text-to-video

fal.aiText-to-videoAudio inputMulti-shot prompt

Generate prompt-only Wan 2.7 clips with aspect ratio, duration, negative prompt, audio URL, and prompt-expansion controls.

Best for: Use this for ideation, ad concepts, product teasers, social clips, and any workflow where you do not need an input frame or source video.

NewFrame-guided

Endpoint

fal.ai Image-to-Video

fal-ai/wan/v2.7/image-to-video

Image-to-videoFirst + last frameVideo continuation720p / 1080p

Animate a starting image, define an ending frame, or continue an existing clip while keeping Wan 2.7's motion and camera logic under one endpoint family.

Best for: Best for product reveals, photo animation, first-and-last-frame transitions, or continuing a source clip with more controlled structure.

NewReference

Endpoint

fal.ai Reference-to-Video

fal-ai/wan/v2.7/reference-to-video

Reference-to-videoMulti-videoCharacter consistencyMulti-shot

Generate new scenes from one or more reference videos and images, with support for multi-subject prompts and optional multi-shot narration.

Best for: Use this when character identity, product appearance, or source-scene consistency matters more than raw one-shot generation speed.

NewEdit

Endpoint

WaveSpeed Video Edit

alibaba/wan-2.7/video-edit

WaveSpeedVideo editReference imagesREST API

Edit an existing video with natural-language instructions, optional 1 to 9 reference images, explicit resolution and duration controls, and audio handling options.

Best for: A good fit for variant creation, recolors, object swaps, scene restyling, and editing a winning clip instead of regenerating it from scratch.

NewRoute

Endpoint

OpenRouter Video Route

POST /api/v1/videos with model alibaba/wan-2.7

OpenRouterUnified APIAsync jobsProvider routing

OpenRouter routes Wan 2.7 through a unified asynchronous video API with normalized prompt, duration, resolution, frame-image, and reference parameters.

Best for: Best when you want one integration surface across multiple video models, provider fallback, or pricing and uptime abstraction at the routing layer.

Get started today

Ready to integrate Wan 2.7?

Try the API directly in the console, or reach out to the team for onboarding, pricing, and enterprise setup.

API Documentation

How to get access to Wan 2.7 API

The provider surfaces are different, but the workflow stays the same: choose prompt-only, frame-guided, reference-guided, or edit mode, submit an async job, then poll until the final video URL is ready.

Selected endpoint

fal-ai/wan/v2.7/text-to-video

Start with text-to-video for fast validation, then move into image-to-video, reference-to-video, or video edit when the workflow needs tighter scene control.

Use this for ideation, ad concepts, product teasers, social clips, and any workflow where you do not need an input frame or source video.

import { fal } from "@fal-ai/client";

fal.config({
  credentials: process.env.FAL_KEY
});

const result = await fal.subscribe("fal-ai/wan/v2.7/text-to-video", {
  input: {
    prompt: "Luxury fragrance bottle on wet black stone, slow orbit camera, soft cinematic haze, sharp specular reflections, premium launch-film look",
    resolution: "1080p",
    aspect_ratio: "16:9",
    duration: 5,
    enable_prompt_expansion: true
  },
  logs: true,
  onQueueUpdate(update) {
    if (update.status === "IN_PROGRESS") {
      update.logs?.forEach((log) => console.log(log.message));
    }
  }
});

console.log(result.data.video.url);

Async flow

  1. 1

    Choose the provider and mode first: prompt-only generation, image-to-video, reference-to-video, or prompt-driven video edit.

  2. 2

    Submit the request with prompt, duration, resolution, and any frame, reference, audio, or edit inputs needed for that workflow.

  3. 3

    Store the returned request or job id, then poll the provider's status endpoint or routed video endpoint until the run is completed.

  4. 4

    Read the finished video URL, usage metadata, and any actual prompt expansion output, then deliver or iterate from the result.

What Makes It Different

What makes the Wan 2.7 API different

This section is laid out to read more like a product narrative than a feature list. Each row shows a capability, why it matters, and what that looks like in a real workflow.

Preview

One family across four video workflows

That gives teams a cleaner migration path from ideation to controlled production instead of switching to a totally different model the moment consistency or editing enters the workflow.

Capability

One family across four video workflows

Wan 2.7 is useful because the stack does not stop at text-to-video. The current provider pages cover prompt-only generation, image-to-video, reference-to-video, and prompt-driven video editing.

That gives teams a cleaner migration path from ideation to controlled production instead of switching to a totally different model the moment consistency or editing enters the workflow.

Example scenario

A growth team starts with text-to-video for moodboarding, moves to image-to-video for product shots, then uses video edit to create campaign variants from the winning clip.

Capability

First-and-last-frame control

WaveSpeed and fal.ai both position Wan 2.7 image-to-video around stronger frame control, including first-frame and last-frame guided motion.

This is especially practical for product transitions, before-and-after reveals, storyboard interpolation, and clips where you care more about the end state than pure freeform generation.

Example scenario

A brand animation flow starts from a still packshot, defines the last frame as the final shelf-ready composition, and lets Wan 2.7 fill in the motion between them.

Preview

First-and-last-frame control

This is especially practical for product transitions, before-and-after reveals, storyboard interpolation, and clips where you care more about the end state than pure freeform generation.

Preview

Reference-driven identity consistency

That makes Wan 2.7 more useful for multi-character scenes, recurring mascots, spokesperson videos, and product-led storytelling than a pure prompt-only model.

Capability

Reference-driven identity consistency

fal.ai and WaveSpeed both describe Wan 2.7 reference-to-video as a way to preserve character, prop, and style consistency from multiple reference videos and images.

That makes Wan 2.7 more useful for multi-character scenes, recurring mascots, spokesperson videos, and product-led storytelling than a pure prompt-only model.

Example scenario

A creative team combines two source videos and one prop reference so the generated scene keeps both speakers recognizable while introducing a new shared environment.

Capability

Audio-aware video generation

Current fal.ai Wan 2.7 docs expose driving audio for text-to-video and image-to-video, while WaveSpeed video edit includes audio handling and OpenRouter normalizes audio generation flags in its video API.

You can line up rhythm, pacing, and background sound logic earlier in the generation workflow instead of bolting every clip onto a separate sound stack from the start.

Example scenario

A music content product submits a short guide track so a lyric-free teaser clip matches the beat and pacing of the campaign audio.

Preview

Audio-aware video generation

You can line up rhythm, pacing, and background sound logic earlier in the generation workflow instead of bolting every clip onto a separate sound stack from the start.

Unified API Platform

Two API tiers for different use cases

Pick the right balance of quality, speed, and cost for your workflow. The section stays data-driven, but the presentation is closer to a clean product comparison table.

Feature
fal.ai Direct
WaveSpeed RESTRecommended
OpenRouter Route
Best forProvider-native Wan 2.7 integrationClear public pricing and broad mode coverageOne routed API across many video models
SpeedDirect async queue flowFast REST workflow with no cold startsAsync video jobs with fallback options
QualityFull coverage for T2V, I2V, and R2VStrong for T2V, I2V, R2V, and video editNormalized request shape over provider abstraction
CostProvider pricingPublished per-second or per-run pricingOpenRouter routed pricing
Recommended useUse fal.ai when you want direct Wan endpoints, the official JS client, file upload helpers, and queue-oriented job handlingStart here when you want straightforward REST calls, visible pricing tables, and Wan 2.7 modes including video edit under one model familyUse OpenRouter when you want /api/v1/videos, normalized parameters, provider routing, and easier switching between Wan, Seedance, Veo, or Sora-style models
API endpointsfal-ai/wan/v2.7/text-to-video, fal-ai/wan/v2.7/image-to-video, fal-ai/wan/v2.7/reference-to-videoalibaba/wan-2.7/text-to-video, alibaba/wan-2.7/image-to-video, alibaba/wan-2.7/reference-to-video, alibaba/wan-2.7/video-editPOST /api/v1/videos, GET /api/v1/videos/{jobId}

Use Cases

Industries using the Wan 2.7 API

This section keeps the same reusable data model, but the presentation is closer to a grid of industry cards instead of long narrative boxes.

Growth teams, brands, and performance marketers

Cinematic ad generation

Turn product launch concepts, brand moments, and social-first campaign prompts into short cinematic clips with more deliberate camera language than generic template video tools.

Wan 2.7's prompt handling, 1080p output, and optional audio support make it a practical fit for ad ideation and polished short-form creative.

Product marketers, social editors, and creative studios

First-and-last-frame transitions

Animate stills into transitions where you control both the opening and closing visual state instead of guessing the motion path from one frame alone.

This is one of Wan 2.7's most practical differentiators for product reveals, before/after sequences, and storyboard interpolation.

Filmmakers, creator tools, and branded content teams

Character-consistent storytelling

Generate scenes with repeated characters or props by supplying reference videos and images that keep identity and appearance more stable across clips.

Reference-to-video is more useful than prompt-only generation when continuity matters across a narrative or campaign.

Ecommerce teams and creative ops

Video variant editing

Edit a winning video into new variants by changing colors, materials, props, or mood without rebuilding the entire clip from scratch.

WaveSpeed's Wan 2.7 video edit path is useful when you want natural-language changes plus optional image guidance while keeping temporal continuity.

Music products, social apps, and AV workflows

Music and audio-led motion

Guide clip rhythm or soundtrack behavior by attaching audio input or using routed audio-generation parameters in providers that support them.

This makes Wan 2.7 stronger for lyric-free teasers, audio-synced mood pieces, or short-form marketing clips that need pacing tied to sound.

Platform teams and multimodel products

Model routing without rewrite risk

Start with a direct provider during experimentation, then route through a unified video API once you need fallback, pricing abstraction, or easier switching across models.

OpenRouter's /api/v1/videos and similar routed layers reduce integration churn when the video model landscape moves faster than your application roadmap.

Examples

Wan 2.7 API examples

Prompt directions paired with visual reference frames. Use them as inspiration for landing pages, creator tooling, commercial mockups, or API playground defaults.

Cinematic atmosphere reference for a Wan 2.7 neon portrait prompt.

Neon rain portrait

Text-to-video mood piece

A strong T2V prompt spells out subject, camera motion, lighting, reflections, and emotional tone instead of relying on generic cinematic keywords.

Close-up portrait of a woman pressing her palm against a rain-soaked window at night, neon reflections in cyan and magenta, shallow depth of field, slow push-in camera, cinematic lens artifacts, emotional stillness.

text-to-videoportraitneon
Lifestyle image used as inspiration for a Wan 2.7 first-and-last-frame prompt.

Kite handoff transition

First-and-last-frame control

This is the kind of structured I2V prompt where Wan 2.7's frame endpoints matter more than freeform camera invention.

Use the first frame as a father and daughter launching a kite in a bright park and the last frame as the kite soaring high above them at golden hour. Fill the motion with a soft circling camera and natural wind-driven movement.

image-to-videofirst framelast frame
Structured narrative reference image for a Wan 2.7 reference-to-video example.

Cafe reunion with references

Reference-to-video scene

Use numbered references in the prompt when your workflow depends on who appears where and how multiple subjects interact in the final shot.

Video 1 and Video 2 meet in a warm neon cafe at dusk. Video 2 places image 3 on the table while Video 1 leans forward smiling. Keep both identities stable, preserve wardrobe details, and add subtle camera drift.

reference-to-videomulti-characterstorytelling
Demo

Motorcycle recolor edit

Prompt-driven video edit

A practical edit prompt isolates the intended change while explicitly preserving motion, framing, and everything that should remain untouched.

Change the motorcycle from matte black to saturated cobalt blue. Preserve motion blur, rider position, camera path, and the urban night lighting already present in the clip.

video editvariantnatural language

How To Use This API

How to use Wan 2.7 API

This quick-start walkthrough is written to rank for integration-style searches while staying concise enough for busy developers and operators.

  1. 1

    Choose your Wan 2.7 mode

    Start by deciding whether the workflow is prompt-only generation, frame-guided motion, reference-driven generation, or editing an existing source clip.

  2. 2

    Pick the provider path

    Use fal.ai for direct Wan endpoints, WaveSpeed when you want visible REST pricing and model coverage, or OpenRouter when you want one async video API across multiple vendors.

  3. 3

    Prepare reference inputs carefully

    Use clean first frames, distinct reference videos, and optional audio only when each input is materially helping motion, identity, or pacing instead of adding noise.

  4. 4

    Set resolution, aspect ratio, and duration

    Run 720p and shorter durations during exploration, then rerender at 1080p or longer duration only after you know the scene direction is working.

  5. 5

    Submit async jobs and poll results

    All three provider paths are asynchronous at the generation layer. Store the request or job id and make result polling part of your normal application flow.

FAQ

Frequently asked questions about Wan 2.7 API

FAQs stay compact and skimmable here. The content is still data-driven for SEO, but the layout is cleaner and less visually heavy.

What is Wan 2.7 API?

Wan 2.7 API is the current provider-hosted or routed interface for Alibaba's Wan 2.7 video generation stack, covering text-to-video, image-to-video, reference-to-video, and, on some providers, prompt-driven video editing.

Does Wan 2.7 support text-to-video and image-to-video?

Yes. fal.ai and WaveSpeed both expose Wan 2.7 text-to-video and image-to-video paths, with support for 720p and 1080p, flexible aspect ratios, negative prompts, and optional prompt expansion.

Can I control both the first and last frame?

Yes. Current Wan 2.7 image-to-video provider pages explicitly support first-and-last-frame workflows, which is useful for transitions, product reveals, and more controlled motion between two visual endpoints.

What is reference-to-video used for?

Reference-to-video is the mode to use when you need character, prop, or scene consistency from one or more source videos and images. Current fal.ai and WaveSpeed pages position it for multi-character scenes and identity-consistent generation.

Does Wan 2.7 support audio-aware generation?

Yes, depending on the provider and mode. fal.ai documents driving audio for text-to-video and image-to-video, WaveSpeed exposes audio handling in video edit, and OpenRouter normalizes a generate_audio flag in its video generation API.

What resolutions and durations are available?

Across the current cited provider pages, Wan 2.7 commonly supports 720p and 1080p. Text-to-video and image-to-video typically run from 2 to 15 seconds, while reference-to-video and video edit are usually shorter-duration workflows.

How much does Wan 2.7 cost?

As of April 24, 2026, OpenRouter lists Wan 2.7 at $0.10 per second, and WaveSpeed lists 720p text-to-video, image-to-video, and video-edit pricing at $0.10 per second, with 1080p at $0.15 per second. WaveSpeed reference-to-video starts at $1.00 for 5 seconds at 720p because reference processing adds overhead.

Which provider should I start with?

Use fal.ai when you want direct Wan 2.7 endpoints and the provider-native client, WaveSpeed when you want straightforward REST pricing and mode coverage, or OpenRouter when you want routed video jobs and easier switching between Wan and other video models.

Why use ImaRouter for Wan 2.7 instead of wiring each provider separately?

ImaRouter combines model routing, five-modality coverage, transparent pricing, automatic failover, and faster new-model onboarding so teams do not have to integrate and monitor providers one by one.

Model Directory

Browse the full model market before you choose your route.

Use the `/models` catalog to scan providers, modalities, reasoning support, context windows, and pricing metadata from a local OpenRouter snapshot. It is the fastest way to compare what exists before you decide which models should be prioritized on ImaRouter.

Get Started

Validate Wan 2.7 on the provider path you trust, then productionize the workflow that wins

Use WaveSpeed for clear REST pricing, fal.ai for direct Wan endpoint control, or OpenRouter for routed video jobs. Once the prompt, frame, or reference workflow is stable, move it into production with the async pattern that best fits your stack.