Gemini Omni MCP

FastMCP server for Google's gemini-omni-flash-preview video model. It supports text-to-video, image-to-video, multiple reference images, uploaded-video edits, stateful edits with previous_interaction_id, URI delivery, inline delivery, and batch generation.

Install

uvx gemini-omni-mcp

Or from source:

uv sync --all-extras
uv run gemini-omni-mcp

Related MCP server: veo-mcp-server

Configuration

export GEMINI_API_KEY=your_key_here
export OUTPUT_DIR=~/gemini_omni_videos

Optional settings include REQUEST_TIMEOUT, FILE_POLL_INTERVAL, FILE_POLL_TIMEOUT, MAX_BATCH_SIZE, DEFAULT_ASPECT_RATIO, DEFAULT_DELIVERY, and DEFAULT_DURATION_SECONDS.

Tools

generate_video

Generates one MP4 and returns JSON with video.path, interaction_id, and metadata.

Important arguments:

• prompt: scene, motion, camera, lighting, mood, and audio direction

• task: text_to_video, image_to_video, reference_to_video, or edit

• aspect_ratio: 16:9 or 9:16

• duration_seconds: optional preview field, 3 to 10

• reference_image_paths: up to 6 local image paths

• input_video_path: local MP4 to upload and edit

• delivery: uri or inline

• previous_interaction_id: continue editing a generated video

batch_generate

Runs multiple prompts in conservative parallel batches, capped at 4.

Prompting tips

• Use "single continuous shot" and "no scene cuts" for one-scene outputs.

• Include audio direction, for example "gentle ambient sound, no dialogue".

• For edits, keep the prompt short and add "Keep everything else the same".

• Use <FIRST_FRAME> and <IMAGE_REF_N> tags to bind reference-image roles.

• Timing cues like [0-3s], [3-6s], and [6-10s] work well.

Development

uv sync --all-extras
uv run ruff format .
uv run ruff check .
uv run mypy gemini_omni_mcp/
uv run pytest
uv build