MCP servers that can list available models

4 verified servers expose a tool that can list available AI models

An agent that calls another model first needs to know which models it can call. A list-models tool returns the catalog a provider offers, so the agent picks the right one for the job instead of hard-coding a name that may have changed. Catalogs move; names go stale.

These verified servers let an agent list the models a provider offers.

Top pick

Google Gemini

Ali Argun

Community

Maintained community MCP server for Google's Gemini API: generate text, analyze images, count tokens, and create embeddings from your agent.

ai-ml255
Tool:
  • list_models

Gemini's list_models returns its models with their capabilities, so an agent can match a task to the right model in the Google family.

Pick 2

fal.ai

Raveen Beemsingh

Community

Community MCP server for fal.ai: generate and edit images, video, music, and audio with 600+ fast generative models from your agent.

ai-ml48
Tool:
  • list_models

fal's list_models discovers the hundreds of models on its platform with filtering, for an agent picking among image, video, and audio models by capability.

Pick 3

Replicate

Replicate

Official

Replicate's official MCP server: discover, compare, and run thousands of hosted AI models — image, video, audio, and language — straight from your agent.

ai-ml
Tool:
  • list_models

Replicate paginates its public models through list_models, the catalog an agent browses to find a community or official model to run.

Pick 4

ElevenLabs

ElevenLabs

Official

ElevenLabs' official MCP server: text-to-speech, voice cloning, speech-to-text, sound effects, and conversational AI agents from your editor.

ai-ml
Tool:
  • list_models

ElevenLabs scopes list_models to its speech-synthesis models, so an agent choosing a voice selects from what the account can actually use.

What to know

Model catalogs move: providers add models, deprecate others, rename versions. An agent that lists them at runtime picks from what is actually available rather than a name baked in months ago. What comes back differs by provider. Gemini returns its models with their capabilities. fal exposes a catalog of hundreds with filtering. Replicate paginates its public models. ElevenLabs lists its speech-synthesis models. The use is the same across them: discover, then call the model that fits the task and the budget.

The set a provider offers is stable enough within a session that an agent need not re-list it on every call. Which model it chose for a kind of task, and why, is the more useful thing to carry forward, so a later run reuses that decision instead of re-deciding from the whole catalog.

Questions

Why list models instead of hard-coding one?
Because catalogs change. Providers add and deprecate models and rename versions, so a name hard-coded months ago may no longer exist. Listing at runtime lets an agent pick from what is currently available and fall back gracefully when a model is gone.
What does the list tell the agent?
Usually the model ids plus some capability or metadata, so an agent can match a task to a model. Gemini returns capabilities; fal supports filtering across a large catalog; Replicate paginates. From the list the agent picks the id it then calls.