Video

gemini-omni-1.1-flash

Google's GA multimodal video model supports generation, natural-language editing, frame interpolation, and video extension. APIMart provides it through a dedicated official channel with asynchronous task tracking and resolution-based billing.

Multimodal VideoFirst-Last Frame360p to 4K

Get API Key API Docs

Service Highlights

99.9% SLA
Official Discounts
Pay-as-you-go
Fast & Low Latency

Input

Model

Prompt

68 characters

Aspect Ratio

Resolution

Task

Prefer automatic inference unless the intended behavior is ambiguous.

First / Last Frame

First Frame

Last Frame

Reference Image

Reference Video

Previous Task ID

Output

Video link valid for 72 hours

Multimodal Video Generation and Editing in One API

Generate from text, animate image references, interpolate between first and last frames, or edit an existing clip. Select 360p, 720p, 1080p, or 4K output and manage every job through APIMart's asynchronous task workflow.

Get API Key

50K+

Active Users

99.9%

Uptime

Faster

70%

Cost Savings

Gemini Omni 1.1 Flash API Capabilities

Documented controls for building generation, interpolation, editing, and extension workflows

Four Resolution Tiers

Choose 360p, 720p, 1080p, or 4K. The default is 720p, while 1080p and 4K are upsampled outputs. The API also accepts 2160p as an alias for 4K.

First and Last Frame Interpolation

Set first_frame_image to animate a starting image, or add last_frame_image to guide the final state. A last frame cannot be submitted by itself, and both frame images should use matching proportions.

Flexible Image References

Provide image_urls for subject references or image_with_roles for explicit first_frame and last_frame roles. Reference images and frame images count toward a combined limit of 10.

Video Editing and Extension

Edit one source video of up to 10 seconds with a concise instruction, or continue a successful APIMart task with extend_from_task_id. Conversational extensions can build a sequence up to approximately 40 seconds.

Video with Generated Audio

Create 24 fps clips with generated dialogue, music, or sound effects when requested in the prompt. Audio files are not accepted as input, so sound direction belongs in the text instruction.

Asynchronous REST Workflow

Submit a job to POST /v1/videos/generations, save the returned task_id, and poll GET /v1/tasks/{task_id}. Stop polling when the status becomes completed or failed.

Automatic Task Selection

Let the service infer the workflow from your inputs, or pass metadata.task as text_to_video, image_to_video, reference_to_video, edit, or extend when explicit intent is useful.

Resolution-Based Billing

Resolution determines the output-token tier. APIMart preauthorizes credits for the maximum 10-second output, then reconciles the final charge against actual usage; failed tasks are not charged.

Practical Multimodal Video Workflows

Use the model's documented input modes for controlled creative and production tasks

First-to-Last Frame Transitions

Define the opening and closing images for product reveals, scene changes, before-and-after sequences, and other transitions that need a clear visual destination.

Reference-Driven Product Motion

Supply product and character images as references, then describe how those subjects should move and interact while keeping the creative brief in one request.

Targeted Footage Revisions

Upload a short clip and request a focused change in natural language. Simple instructions such as changing one object while keeping everything else the same work best for editing.

Campaign Storyboards

Test camera direction, pacing, mood, lighting, and audio ideas at 360p or 720p before committing credits and production time to a higher-resolution final render.

Landscape and Vertical Content

Generate 16:9 assets for presentations and web video or 9:16 assets for mobile-first campaigns. Unsupported aspect ratios fall back to the 16:9 default.

Automated Video Pipelines

Integrate submission, status polling, cost reconciliation, and result retrieval into internal creative tools without maintaining a separate provider-specific task system.

Start Generating with the API

Create a key, submit a valid asynchronous request, and retrieve the completed video

Create an APIMart API Key

Sign in to APIMart, create an API key, and send it as a Bearer token in the Authorization header. Keep the key on your server and never expose it in public client code.

→

Build a Valid Request

Set model to gemini-omni-1.1-flash and provide a prompt or at least one supported image or video asset. If omitted, resolution defaults to 720p and aspect_ratio defaults to 16:9.

→

Submit, Poll, and Save the Result

Submit the generation request, poll the returned task_id until completion, and read the hosted video URL from result.videos[0].url. Download results you need to retain long term.

Generate Your First Video

Gemini Omni 1.1 Flash API FAQ

Supported inputs, parameters, limits, billing, and task handling on APIMart

What is Gemini Omni 1.1 Flash API?

It is Google's GA multimodal video model for text-to-video, image-to-video, first-last frame interpolation, video editing, and conversational extension. On APIMart, use the dedicated model ID gemini-omni-1.1-flash with the unified video endpoint.

Which request parameters are supported?

The request supports model, prompt, resolution, aspect_ratio, image_urls, first_frame_image, last_frame_image, image_with_roles, video_urls, metadata.task, and extend_from_task_id. A prompt or at least one image or video asset is required.

Can I set the video duration?

No. Generated clips are approximately 3–10 seconds long, but this API does not expose a duration parameter. Output is 24 fps and includes generated audio.

How do first and last frame inputs work?

Use first_frame_image by itself for image-to-video, or submit it together with last_frame_image for interpolation. Submitting only last_frame_image returns a 400 invalid_frame_images error. The two images should have the same proportions and match aspect_ratio.

What image, video, and audio inputs can I provide?

You can provide up to 10 images in total, including first and last frames, plus no more than one reference or editable video of up to 10 seconds. Public HTTP(S) URLs and supported base64 data URLs are accepted. Audio input and YouTube links are not supported.

How do I submit and monitor a generation task?

Send POST /v1/videos/generations and store the returned task_id. Poll GET /v1/tasks/{task_id} every 5–10 seconds until the task is completed or failed. Allow a client timeout of up to 10 minutes because 4K and video-input jobs take longer.

How do editing and conversational extension differ?

For direct editing, pass one clip in video_urls with a focused prompt. To continue a prior result without uploading it again, pass the successful APIMart task ID in extend_from_task_id. The previous task must belong to the current user and use an Omni model.

How does resolution-based billing work?

The selected resolution determines the output-token rate. APIMart first reserves enough credits for a maximum 10-second result, then adjusts the charge to the actual token usage after success. A failed task is not charged, and insufficient reserved balance returns HTTP 402.

Which common generation controls are not supported?

The endpoint does not accept system prompts, temperature, a negative-prompt field, audio input, or a duration parameter. Put exclusions and audio direction directly in prompt instead.

How can I contact support for help?

You can reach us via the live chat in the bottom-right corner, email us at [email protected], or join our Discord community. Our team will get back to you as soon as possible.

Related Models

Explore more models in the same category.

View All Models →