Drop files here, paste files, or add a URL. Use no image for text-to-video, one image for image-to-video, or three images for reference fusion.
Drop files here, paste files, or add a URL. video/mp4,video/quicktime,video/webm,video/x-matroska · 100 MiB
Example Output
This is sample data. Generate a new result to see real output.
Example video
Transparent pricing with no hidden fees. Pay as you go.
omni-flashomni-flashomni-flashomni-flashomni-flashomni-flashomni-flashomni-flashomni-flashomni-flash| Model | Spec | PoYo price | Official price | You save |
|---|---|---|---|---|
| omni-flash | 720p/1080p - no video input - 4s | $0.600/gen 120 credits /gen | - | - |
720p/1080p - no video input - 6s | $0.750/gen 150 credits /gen | - | - | |
720p/1080p - no video input - 8s | $1.00/gen 200 credits /gen | - | - | |
720p/1080p - no video input - 10s | $1.10/gen 220 credits /gen | - | - | |
4K - no video input - 4s | $1.25/gen 250 credits /gen | - | - | |
4K - no video input - 6s | $1.50/gen 300 credits /gen | - | - | |
4K - no video input - 8s | $1.75/gen 350 credits /gen | - | - | |
4K - no video input - 10s | $2.25/gen 450 credits /gen | - | - | |
720p/1080p / with video input | $1.50/gen 300 credits /gen | - | - | |
4K / with video input | $2.00/gen 400 credits /gen | - | - |
* Actual fees are based on the final output.
Complete guide to using Omni Flash API for Multimodal Video Generation
The page examples use PoYo-hosted CDN media and are matched to the supported API modes: text-to-video, image-to-video, and video-input generation.
01
Use text-only generation when you want fast concept clips without preparing source media. Describe scene physics, camera movement, material behavior, and atmosphere in the prompt, then choose 720p or 1080p output.
02
For image-to-video, upload one source image and explain how it should move. This official Google DeepMind dandelion drawing example keeps the simple line art as the visual anchor while turning it into a short animated scene.
03
When a source video is provided, Omni Flash switches to a video-input workflow. Use it to transfer motion cues, timing, or scene structure while optionally adding reference imagery through the same request.
Generate short clips from plain prompts for pitches, storyboards, and creative direction tests.
Animate product photos, illustrations, and still frames into short social or campaign clips.
Submit three source images when a scene needs multiple visual references in one generation request.
Use a video reference to guide motion while adjusting the final look and output resolution.
| Feature | Omni Flash | Veo 3.1 | Seedance 2.0 |
|---|---|---|---|
| Provider | Google Gemini Omni | Google DeepMind Veo | ByteDance Seedance |
| Best For | Conversational video editing, prompt-to-video, image-to-video, and reference-guided iteration | Cinematic generation, native audio, image-guided shots, and production previsualization | Multi-shot stories, native audio-video generation, lip-sync, and complex multimodal references |
| Official Positioning | Create and edit video from any input, with world understanding and conversational refinement | Leading video generation model for filmmakers and storytellers, with realism, prompt adherence, and audio | Native multimodal audio-video generation model for text, image, audio, and video inputs |
| Input Modes on PoYo | Prompt only, 1 image, 3 images, or 1 video URL | Prompt, image guidance, first/last frames, reference images, or video extension depending on tier | Text, image, video, and audio reference workflows through the Seedance 2 API page |
| Output and Audio | Video output on PoYo; official model card describes high-resolution video with audio | Video with native audio or silent output depending on model settings | Native audio-video joint generation with dialogue, SFX, music, and ambience |
| Resolution on PoYo | 720p, 1080p | 720p or 1080p depending on tier | 480p, 720p, 1080p depending on model |
| Duration on PoYo | 4s, 6s, 8s, 10s for no-video-input jobs; video input uses its own mode | Usually 4s, 6s, or 8s generation; extension supports source-video continuation | 4-15 seconds, billed per second |
| Published Evaluation Status | Google says detailed T2VA, I2VA, R2VA, editing, and image generation evaluations will be shared when developer and enterprise APIs roll out | Google publishes Veo 3.1 preference, alignment, visual quality, audio-video alignment, and physics claims | Seedance 2.0 model card reports broad improvements and leading-level performance in expert and public user tests |
| Choose This When | You need Google Omni Flash search coverage, conversational editing language, and compact PoYo API controls | You need a mature cinematic model with native audio options and stronger public benchmark evidence | You need longer multi-shot clips, audio/lip-sync workflows, and richer multimodal references |
Model capabilities change quickly. The table combines public model information and PoYo's current API surface as of June 2026; verify current request fields and pricing in the PoYo docs before production integration.
Step 1: Create your API key
Generate a PoYo API key and submit requests to the unified asynchronous generation endpoint.
Create API Key
Step 2: Choose an input mode
Send only prompt for text-to-video, add one image_urls item for image-to-video, send three images for reference fusion, or add one video_urls item for video-input generation.
Add Credits
Step 3: Poll status or use webhook
Use the returned task ID to query progress, or provide callback_url to receive the finished video asynchronously.
Open API Documentation
model: "omni-flash" with a prompt and optional image_urls or video_urls, then poll the status endpoint or use a webhook to retrieve the video.Submit tasks through the same PoYo generation endpoint and retrieve final files with the standard task status flow.
Resolution, duration, aspect ratio, image count, and video input are explicit request fields.
The model card and pricing table expose duration-based and video-input credit tiers before generation.
Task input records keep user-facing fields only, which simplifies debugging and customer support.