Generate 15-Second Videos with Gemini Omni 1.2 Flash API

Google released Gemini Omni 1.2 Flash on November 5, 2026. The model generates and edits videos up to 15 seconds long with native synchronized audio. It costs $0.07 per second for 720p output, a 30% reduction from Omni 1.1 Flash.

Developers access it through the Gemini API Interactions endpoint as gemini-omni-1.2-flash. You can draft at 360p or 720p, edit conversationally, and upscale the best results to 1080p or 4K.

This guide shows how to generate your first 15-second video, control duration and resolution, and perform multi-turn edits. For product-focused examples, see Create 15s Product Videos with Gemini Omni 1.2 Flash.

Set Up the Gemini API

Create a Google AI Studio API key and enable billing. Install the current google-genai Python package.

pip install -U google-genai

Set your key as an environment variable:

export GEMINI_API_KEY="your-key-here"

The Interactions API returns video data as base64 by default. For longer clips, request URI delivery to avoid payload limits.

Generate a Text-to-Video Clip

Start with a simple prompt. Specify duration in the response format for exact length control. Durations range from 3 to 15 seconds in one-second steps. The default is 10 seconds.

import base64
from google import genai

client = genai.Client()

interaction = client.interactions.create(
    model="gemini-omni-1.2-flash",
    input="A continuous tracking shot of a golden retriever running through a field of wildflowers at golden hour. Soft ambient sounds of wind and distant birds. No dialogue.",
    response_format={
        "type": "video",
        "duration": "15s",
        "resolution": "720p",
        "aspect_ratio": "16:9"
    }
)

with open("retriever.mp4", "wb") as f:
    f.write(base64.b64decode(interaction.output_video.data))

The model produces synchronized audio automatically. Describe the sound design in your prompt for better results.

Control Resolution and Aspect Ratio

Supported resolutions are 360p, 720p (default), 1080p (upscaled), and 4k (upscaled). Aspect ratios are 16:9 or 9:16.

Use 360p for rapid iteration. It generates roughly twice as fast and consumes fewer tokens. Upscale only the clips you keep.

First-and-last-frame interpolation supports 3 to 10 seconds. Reference images (up to six) work across all durations.

Multi-Turn Conversational Editing

Pass the previous interaction ID to continue editing. The model preserves unmentioned elements while applying your changes.

# Continue from the previous interaction
edit = client.interactions.create(
    model="gemini-omni-1.2-flash",
    previous_interaction_id=interaction.id,
    input="Change the field to a snowy mountain trail. Keep the dog and camera movement identical. Add soft snow crunching sounds."
)

with open("retriever_snow.mp4", "wb") as f:
    f.write(base64.b64decode(edit.output_video.data))

You can extend clips by 3 to 10 seconds per turn, up to a total of 30 seconds. Prompt for continuity: “The scene continues as the dog reaches the summit.”

Use Reference Images

Upload or pass image URIs as reference inputs. The model maintains character, product, and style consistency across the clip.

Combine a character sheet with a product photo for virtual try-on sequences or branded explainers. Reference videos up to three seconds work for motion guidance on longer outputs.

Tips for Better Results

Write prompts with six layers: subject, action, setting, camera, lighting, and audio. Request a single continuous shot when you need no cuts.

Use timecode syntax for multi-beat stories:

[0-5s] Close-up of the product on a wooden desk in soft morning light
[5-10s] A hand lifts it and pours water into a glass
[10-15s] The product sits on a lakeside rock as the camera orbits

Test at 360p first. Check lip-sync and physics on the draft before upscaling. Store the interaction ID if you plan further edits.

Avoid asking for complex multi-character dialogue in one generation. Break long sequences into editable turns.

Conclusion

Gemini Omni 1.2 Flash makes 15-second video production practical for developers. Generate a draft, edit it conversationally, and upscale the final version without regenerating from scratch. Start with the Interactions API examples above and refine prompts against the official documentation.

Sources

• Introducing Gemini Omni 1.2 Flash
• Generate and edit videos with Gemini Omni Flash
• Gemini Omni 1.2 Flash on Google Cloud