LLM Gateway MCP server

• Shows who you are connected as

  Shows the connected user, organization, project, role and any API key spending limits. Owners and admins also see the organization's credit balance, and credentials are never returned.

  Try: “Which LLM Gateway project and role am I connected as?”

  Read · get-account

• Shows your usage and costs

  Returns request and token totals, errors, cache hits, costs and a time series for a date range, plus your most-used provider, model and coding app.

  Try: “Show my usage and costs for the last 30 days.”

  Read · get-usage

• Ranks providers, models, apps or keys

  Ranks providers, models, coding apps or API keys by requests, cost or tokens, with paging.

  Try: “Which coding app accounts for the most cost this month?”

  Read · get-usage-breakdown

• Lists available LLM models

  Lists models with capabilities and pricing, with filters for family and deprecated models.

  Try: “List the Anthropic models you can reach, with prices.”

  Read · list-models

• Lists image models

  Lists the image generation models you can use.

  Try: “Which image models are available?”

  Read · list-image-models

• Sends a message to any model

  Sends a conversation to a model you choose and returns the full answer. The cost is billed to the workspace that issued the key.

  Try: “Ask a different model to review this function.”

  Write · chat

• Generates an image from a prompt

  Generates one to four images from a text prompt with an image model, defaulting to qwen-image-3.0. Image generation is billed to your workspace.

  Try: “Generate a minimalist blue logo for my startup.”

  Write · generate-image

• Generates an image with Gemini 3 Pro

  Generates an image with Gemini 3 Pro Image and returns an inline preview. The hosted server returns it inline without writing files, since saving needs a server upload directory.

  Try: “Create a 16:9 hero image for my landing page.”

  Write · generate-nano-banana

Watch out for

• Chat and image tools spend your credits

  chat, generate-image and generate-nano-banana are billed to the workspace that issued the key. Spending limits on the key or member still apply to them. docs.llmgateway.io

• The API key sits in your client config

  The key is sent as a Bearer header, so it is stored in the config file. Use a key limited to one project and rotate it if the file is shared. docs.llmgateway.io

• Usage tools only read your own scope

  Owners and admins see the project's usage across keys, and developers see only keys they created. The tools cannot override scope. docs.llmgateway.io

Security advisories