Gemini MCP runs on your own machine — the client starts it, so there is no endpoint to ping. 1 893 installs a week from npm. Last commit 1 Nov 2026.
Generate and edit images with Google Gemini (Nano Banana / Nano Banana Pro) image models.
Today is the operative word: we check Gemini MCP every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add gemini-mcp -- npx -y @chrischall/gemini-mcp
{
"mcpServers": {
"gemini-mcp": {
"command": "npx",
"args": [
"-y",
"@chrischall/gemini-mcp"
]
}
}
}
[mcp_servers.gemini-mcp]
command = "npx"
args = ["-y", "@chrischall/gemini-mcp"]
{
"mcpServers": {
"gemini-mcp": {
"command": "npx",
"args": [
"-y",
"@chrischall/gemini-mcp"
]
}
}
}
{
"servers": {
"gemini-mcp": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@chrischall/gemini-mcp"
]
}
}
}
Optional settings from the registry entry, left out of the block above:
GEMINI_API_KEY (Your Google Gemini API key (aistudio.google.com/apikey)), GEMINI_IMAGE_MODEL (Override the default image model (default: gemini-nano-banana-2.1)), GEMINI_OUTPUT_DIR (Default directory for generated images (default: ~/Downloads/gemini-mcp). When set, a per-call output_dir must be inside it), GEMINI_INPUT_DIR (Directory to resolve bare input-image filenames against), GEMINI_UPLOAD_DIR (Restrict local-file uploads to the Gemini Files API to these directories (path.delimiter-separated; unset = unrestricted)), GEMINI_REFERENCE_DIR (Restrict local reference images (images / image paths sent to Gemini) to these directories (path.delimiter-separated), plus GEMINI_OUTPUT_DIR. Unset: GEMINI_UPLOAD_DIR applies; if that is unset too, unrestricted locally, and $MCP_DATA_DIR/uploads when MCP_DATA_DIR is set (hosted)), GEMINI_TIMEOUT_MS (Upstream request timeout in ms (default: 60000, or 120000 for image_size: "4K"); each generation tool also takes a per-call timeout_ms), GEMINI_HEARTBEAT_MS (Progress-notification cadence in ms while a generation runs (default: 10000; 0 disables) — keeps MCP hosts that reset their timeout on progress from timing out long generations), GEMINI_CHAIN_RETRY_MS (How long to wait out interactions-store lag when a chained call 404s (default: 120000; 0 disables retrying)), GEMINI_DEBUG (When set, emit heartbeat diagnostics to stderr. Off by default), GEMINI_RATE_CARD (Override/extend the shipped USD-per-1M-token rates used by gemini_token_usage's cost estimate, e.g. {"gemini-3-pro-image":{"input":2,"text_output":12,"image_output":120}}. Malformed JSON is ignored (a bad card must not fail generations)).
AI image generation and editing with Google Gemini: Nano Banana 2.1, Pro and Lite models
Generate images (Nano Banana), video (Veo, Gemini Omni), speech and music (Lyria) with Google models
Nano Banana MCP server — Google Gemini image generation and editing via Model Context Protocol
Nano Banana Pro AI image editor free online. Create, edit, and enhance images by Google Gemini AI
MCP server for Google Nano Banana (Gemini) image generation
AI image generation and editing with Google Gemini. Structured JSON editing.
Google Gemini image generation, editing, and local processing via MCP
MCP server for Google Gemini image generation and editing
Answers built from our own checks of this server.