Gemini Video Generation

Migrate GeminiTools to Veo 3.1 on Vertex AI and save the returned MP4 correctly.

Use Veo 3.1 through GeminiTools on Vertex AI and decode the returned base64 content before saving the MP4.

Google discontinued the example’s default veo-2.0-generate-001 endpoint after June 30, 2026 and recommends veo-3.1-generate-001. The source also converts video.content to the string representation of a bytes object, which corrupts the saved MP4. Apply both corrections below. See Vertex AI release notes.

gemini_video_generation.py

"""Example: Using the GeminiTools Toolkit for Video Generation

An Agent using the Gemini video generation tool.

Video generation only works with Vertex AI.
Make sure you have set the GOOGLE_CLOUD_PROJECT and GOOGLE_CLOUD_LOCATION environment variables.

Example prompts to try:
- "Generate a 5-second video of a kitten playing a piano"
- "Create a short looping animation of a neon city skyline at dusk"

Run `uv pip install google-genai agno` to install the necessary dependencies.
"""

from agno.agent import Agent
from agno.models.openai import OpenAIChat
from agno.tools.models.gemini import GeminiTools
from agno.utils.media import save_base64_data

# ---------------------------------------------------------------------------
# Create Agent
# ---------------------------------------------------------------------------


agent = Agent(
    model=OpenAIChat(id="gpt-4o"),
    tools=[GeminiTools(vertexai=True)],  # Video Generation only works on VertexAI mode
    debug_mode=True,
)

# ---------------------------------------------------------------------------
# Run Agent
# ---------------------------------------------------------------------------
if __name__ == "__main__":
    agent.print_response(
        "create a video of a cat driving at top speed",
    )
    response = agent.get_last_run_output()
    if response and response.videos:
        for video in response.videos:
            if video.content:
                save_base64_data(
                    base64_data=str(video.content),
                    output_path=f"tmp/cat_driving_{video.id}.mp4",
                )

Run the Example

1. Set up your virtual environment

   Mac

   uv venv --python 3.12
   source .venv/bin/activate
   

   uv venv --python 3.12
   .venv\Scripts\activate
   
2. Install dependencies

   uv pip install -U agno google-genai openai
   
3. Export environment variables

   Mac/Linux

   export GOOGLE_CLOUD_LOCATION="your_google_cloud_location_here"
   export GOOGLE_CLOUD_PROJECT="your_google_cloud_project_here"
   export OPENAI_API_KEY="your_openai_api_key_here"
   

   $Env:GOOGLE_CLOUD_LOCATION="your_google_cloud_location_here"
   $Env:GOOGLE_CLOUD_PROJECT="your_google_cloud_project_here"
   $Env:OPENAI_API_KEY="your_openai_api_key_here"
   
4. Authenticate with Google Cloud

   Sign in with Application Default Credentials:

   gcloud auth application-default login
   
5. Use Veo 3.1

   Replace GeminiTools(vertexai=True) with GeminiTools(vertexai=True, video_generation_model="veo-3.1-generate-001", enable_generate_image=False) in the saved file.

6. Decode the returned content

   Replace base64_data=str(video.content) with base64_data=video.content.decode("utf-8") so save_base64_data() receives the base64 string instead of a bytes representation.

7. Run the example

   Save the code above as gemini_video_generation.py, then run:

   python gemini_video_generation.py
   

Full source: cookbook/91_tools/models/gemini_video_generation.py