Gemini

Gemini Omni Flash API

Video API playgroundHOT

Use this page to test the site's Gemini Omni Flash access layer with the actual request schema exposed by the selected model.

Edit the form or JSON payload, submit a real generation task, inspect field requirements, and open the generated Markdown documentation when you are ready to integrate.

Model:
Pricing

Current estimate starts at $0.15/image

Active model
Gemini Omni Flash Lite

720p: 30 credits - Text/Image to Video - Fixed per generation

1080p: 30 credits - Text/Image to Video - Fixed per generation

720p: 34 credits - Text/Image to Video - Fixed per generation

1080p: 34 credits - Text/Image to Video - Fixed per generation

720p: 36 credits - Text/Image to Video - Fixed per generation

1080p: 36 credits - Text/Image to Video - Fixed per generation

720p: 40 credits - Text/Image to Video - Fixed per generation

1080p: 40 credits - Text/Image to Video - Fixed per generation

4k: 60 credits - Text/Image to Video - Fixed per generation

4k: 68 credits - Text/Image to Video - Fixed per generation

4k: 72 credits - Text/Image to Video - Fixed per generation

4k: 80 credits - Text/Image to Video - Fixed per generation

Displayed estimates follow the site's current credit configuration. Provider pricing, account plans, duration, resolution, retries, and failed tasks can change the final cost.

New tasks appear here after you submit.
Request schema

Inspect the active Flash payload

The request sample and field reference update from the same runtime model schema used by the playground.

Request JSON
{
  "model": "Omni-Flash-Ext",
  "prompt": "a girl is dancing happily in a sunny garden",
  "duration": 10,
  "resolution": "1080p",
  "aspect_ratio": "9:16"
}

Use Gemini Omni Flash Lite in a generation request

The playground, request example, fields, and pricing rows on this page use the active Flash model configuration.

Open full API docs

Review authentication, task creation, polling, fields, examples, pricing rows, and common errors. /models/omni-flash-ext.md

promptstringVideo content description. Describe the scene, subject, action, environment, camera movement, visual style, and audio cues in detail.
image_urlsarrayReference image URL array. Omit or pass an empty array for text-to-video, pass 1 image for image-to-video, or pass 3 images for reference image fusion. Two images are not supported.
video_urlsarrayReference video URL array. Omit it or provide one publicly accessible HTTP/HTTPS video URL. Do not pass video_urls together with duration.
durationenumVideo duration in seconds. Supported values are 4, 6, 8, and 10. Do not pass duration together with video_urls.
resolutionenumVideo resolution. Values are case-insensitive. Supported values are 720p, 1080p, and 4k. (720p, 1080p, 4k)
aspect_ratioenumVideo aspect ratio. Common values are 16:9 for landscape and 9:16 for portrait. (16:9, 9:16)
generation_typeenumGeneration type for image inputs. Use frame for first-frame mode with one image, or reference for one or three reference images. (frame, reference)
Capabilities

A complete path from test to integration

Validate the creative result and the production request before wiring the API into a customer workflow.

Working video playground

Submit prompt-driven short video tasks through the same execution path used elsewhere on the site.

Editable JSON payload

Switch between form and API modes to understand how visible controls map into the request body.

Runtime field reference

Inspect field names, accepted types, enum values, defaults, and descriptions from the active schema.

Cost-aware requests

Review credit estimates before submission and apply task, user, and account limits in production.

Integration path

Test, authenticate, poll, and deliver

Start with the smallest useful request, then add production controls around the successful payload.

Validate a prompt

Run a short creative brief in the playground and confirm the output direction before writing integration code.

Authenticate the request

Create an API key, keep it on the server, and send it as a bearer token with the task request.

Poll the task

Store the returned task identifier, check status with bounded polling, and surface clear queued, running, failed, and completed states.

Use cases

Where the Flash API fits

Use the API for repeatable short-video workflows that need structured requests and budget enforcement.

Product ad concept generation
Short social video variants
Prompt and parameter testing tools
Creator workflow automation
Storyboard and pre-visualization services
Batch jobs with explicit cost limits
Community showcase

Successful Flash generations

Open public results when available to review prompts, settings, and completed media.

View showcase
videoOmni-Flash-Ext

Omni-Flash-Ext Video Generation

Create a 6-second insane phonk Minecraft edit. Start with the realistic classroom shot, slowly push toward the laptop while a dark Brazilian/drift phonk beat builds with distorted 808s and sharp cowbells. Add subtle camera movement, film grain, bass vibration, and tension. At exactly ~2 seconds, hit a HUGE 808 drop: rapidly zoom into the laptop and make the screen glitch, crack, flash white, then transform seamlessly into Minecraft. From 2–5 seconds, unleash an extremely fast Minecraft montage with 3–5 rapid gameplay shots, velocity ramps, speed changes, FOV punches, aggressive camera shakes, motion blur, RGB split, glitch frames, screen flashes, particles, and cuts perfectly synced to every kick and cowbell. At 5 seconds, freeze-frame on the strongest moment for a split second, briefly mute the beat, then hit one final massive bass drop with an extreme zoom and glitch transition back to the classroom. Keep the person realistic and consistent. No text.

Sep 24, 2026View details
videoOmni-Flash-Ext

Omni-Flash-Ext Video Generation

Drop a giant orange on Puppycorn! (Create a 16:9 Unikitty! on Cartoon Network Animation FIXED Style Video In 2D and bedroom!)

Sep 24, 2026View details
videoOmni-Flash-Ext

Omni-Flash-Ext Video Generation

Drop 1 Million Cartoon Network logo cubes on Puppycorn! (Create a 16:9 Unikitty! on Cartoon Network Animation FIXED Style Video In 2D and bedroom!) (Why can when he moves away with he get things drop down!)

Sep 24, 2026View details
videoOmni-Flash-Ext

Omni-Flash-Ext Video Generation

Drop 10 Million pears on Puppycorn! (Create a 16:9 Unikitty! on Cartoon Network Animation FIXED Style Video In 2D and bedroom!)

Sep 24, 2026View details
videoOmni-Flash-Ext

Omni-Flash-Ext Video Generation

I challenge you to get 20 trillion Llamas to run over Chase! (Create a 16:9 Paw Patrol Animation FIXED Style Video In 3D and outside!)

Sep 24, 2026View details
videoOmni-Flash-Ext

Omni-Flash-Ext Video Generation

Drop 1 million marshmallows on Chase! (Create a 16:9 Paw Patrol Animation FIXED Style Video In 3D and outside!)

Sep 24, 2026View details
Production safeguards

Protect cost, reliability, and user trust

Server-side keys

Never expose secret API keys in browser code. Route requests through a server action or protected endpoint.

Bounded retries

Use a retry limit, exponential backoff, idempotency, and clear refund rules for provider-side failures.

Provenance preserved

Keep applicable AI labels and generation metadata with the delivered media. Do not build a provenance-removal flow.

FAQ

Gemini Omni Flash API questions

Can I send a real task from this page?

Yes. Sign in, provide the required fields, review the estimated credits, and submit through the embedded playground.

Where does the request example come from?

The page builds the example and field reference from the same runtime catalog entry used by the selected Flash model.

How do I retrieve a result?

Create the task, store its returned identifier, and poll the task endpoint until it completes or fails. The Markdown docs include the current endpoint examples.

How should I handle retries?

Retry only transient provider or network failures, use backoff and idempotency, and stop after a small bounded number of attempts.

How is pricing calculated?

The site maps the active model pricing to credits. Use the pricing page to model duration, batch size, and retries, then verify current account pricing before launch.

Is this the official Google endpoint?

No. This page documents and tests the site's independent API access layer. Check official Google documentation for current first-party endpoints and terms.

Start integrating

Move a tested Flash request into production

Validate the prompt, copy the request structure, keep the key on the server, and add task and budget controls.

Live
playground
JSON
request mode
Markdown
generated docs