Is OpenRouter more expensive than going direct?

Marginally — OpenRouter takes a small margin on top of provider rates. The trade is one credential, instant access to new models, and zero direct-billing relationships.

Does it support Anthropic prompt caching?

Partially. Some provider-specific features pass through; others are abstracted away. For maximum Claude-specific tuning, go direct.

Can I see per-model cost?

Yes. Every prediction returns token counts and the dollar cost, so the agent can keep a running budget across model calls.

OpenRouter

Name: Top MCPs — curated MCP directory
Creator: Top MCPs
License: https://creativecommons.org/licenses/by/4.0/

Route one prompt across 300+ LLMs with a single API key.

Score 68(?)CommunityMIT 64Verified May 26, 2026Top MCPs for AI & Machine Learning →

Install View on GitHub

Quick answer

What it does

Exposes OpenRouter's OpenAI-compatible chat-completions endpoint as MCP tools. List available models, query a specific model, and stream responses, all with cost tracking.

Best for

Multi-model evaluation
Provider redundancy
Accessing long-tail models
Cost-tracking across providers

Not for

Highest-volume production
Workflows that depend on provider-specific features (Anthropic prompt caching, etc.)

Setup recipe

Pick your client, then follow the three steps.

Install

claude_desktop_config.jsonjson

{
  "mcpServers": {
    "openrouter": {
      "command": "npx",
      "args": [
        "-y",
        "openrouter-mcp"
      ],
      "env": {
        "OPENROUTER_API_KEY": "${OPENROUTER_API_KEY}"
      }
    }
  }
}

Paste under mcpServers. Fully quit and reopen Claude after editing.

CLI or .mcp.jsonshell

# export OPENROUTER_API_KEY=YOUR_API_KEY
claude mcp add openrouter -- npx -y openrouter-mcp

Run from your repo. Commit .mcp.json to share with your team.

.cursor/mcp.jsonjson

{
  "mcpServers": {
    "openrouter": {
      "command": "npx",
      "args": [
        "-y",
        "openrouter-mcp"
      ],
      "env": {
        "OPENROUTER_API_KEY": "${OPENROUTER_API_KEY}"
      }
    }
  }
}

Global path: ~/.cursor/mcp.json. Reload window after editing.

.vscode/mcp.jsonjsonc

{
  "servers": {
    "openrouter": {
      "command": "npx",
      "args": [
        "-y",
        "openrouter-mcp"
      ],
      "env": {
        "OPENROUTER_API_KEY": "${OPENROUTER_API_KEY}"
      }
    }
  }
}

VS Code uses the "servers" key (not "mcpServers").

~/.codeium/windsurf/mcp_config.jsonjson

{
  "mcpServers": {
    "openrouter": {
      "command": "npx",
      "args": [
        "-y",
        "openrouter-mcp"
      ],
      "env": {
        "OPENROUTER_API_KEY": "${OPENROUTER_API_KEY}"
      }
    }
  }
}

Open via Cascade → hammer icon → Configure.

cline_mcp_settings.jsonjson

{
  "mcpServers": {
    "openrouter": {
      "command": "npx",
      "args": [
        "-y",
        "openrouter-mcp"
      ],
      "env": {
        "OPENROUTER_API_KEY": "${OPENROUTER_API_KEY}"
      }
    }
  }
}

Open via the Cline sidebar → MCP Servers → Edit.

~/.continue/config.jsonjson

{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "transport": {
          "type": "stdio",
          "command": "npx",
          "args": [
            "-y",
            "openrouter-mcp"
          ],
          "env": {
            "OPENROUTER_API_KEY": "${OPENROUTER_API_KEY}"
          }
        }
      }
    ]
  }
}

Continue uses modelContextProtocolServers with a transport block.

~/.codex/config.tomlshell

# ~/.codex/config.toml
[mcp_servers.openrouter]
command = "npx"
args = [
  "-y",
  "openrouter-mcp",
]
env = { OPENROUTER_API_KEY = "${OPENROUTER_API_KEY}" }

Codex uses TOML. Each server is a [mcp_servers.<name>] subtable.

~/.config/zed/settings.jsonjsonc

{
  "context_servers": {
    "openrouter": {
      "command": {
        "path": "npx",
        "args": [
          "-y",
          "openrouter-mcp"
        ]
      },
      "env": {
        "OPENROUTER_API_KEY": "${OPENROUTER_API_KEY}"
      }
    }
  }
}

Zed calls them "context_servers". Settings live-reload on save.

ChatGPT → Apps directorynone

OpenRouter doesn't ship a hosted HTTPS endpoint today. ChatGPT supports remote MCP servers only — to use this server in ChatGPT you'll need to deploy it to a public HTTPS URL first (e.g. via Cloudflare Workers or Vercel) or wait for an official remote build.

2
Set required secrets
Set OPENROUTER_API_KEY in your shell environment before launching your MCP client.
3
Try a minimum working prompt
Minimum working prompt pending verification. Try any prompt from the MCP’s README once installed.

Tools & permissions

Tools list pending verification. The server exposes tools over MCP; we haven’t yet parsed its capability manifest into this page. Check the GitHub repo for the authoritative list.

Security & scope

Access scope: Network
Sandbox: Prompts and responses flow through OpenRouter's infrastructure. OpenRouter's privacy policy applies.
Gotchas: Sensitive prompts hit OpenRouter logs — check their data retention policy if compliance matters.
API key is account-scoped; one key per environment is good hygiene.

Agent prompt pack

— copy into Claude, Cursor, or ChatGPT.

Paste into Claude, Cursor, or ChatGPT. Edit the [brackets] before sending.

Recommend the best MCP servers for [task: e.g. ai & machine learning work] in [client: Claude].

Constraints:
- Prefer tools that are [official | open-source | read-only] — pick what matters for my use case.
- Exclude MCPs that require [e.g. a paid plan, OAuth-only flows, remote-only transport].
- Return at most 3 picks, ranked.

For each pick include:
1. One-sentence rationale.
2. The ready-to-paste install snippet for my client.
3. Any required secrets I need to create before installing.

Cross-check the top-mcps.com listing: https://top-mcps.com/top-mcps-for-ai-machine-learning

Compare OpenRouter against a real alternative. Swap the second MCP in [brackets] if you want a different match.

Compare OpenRouter MCP vs [Replicate MCP] for the following job: [describe the job, e.g. "let an agent create GitHub issues on bug triage"].

Judge them on:
- Setup time and complexity (what a new user hits first).
- Auth model (none / API key / OAuth 2.1) and credential risk.
- Transport (stdio / Streamable HTTP / SSE) and where the server runs.
- Required secrets and the blast radius if they leak.
- Operational risk in an unattended agent loop.
- Which one is "good enough" for a weekend prototype vs. production.

End with one sentence: which should I pick for my scenario, which is: [my scenario].

References:
- https://top-mcps.com/mcp/openrouter
- top-mcps.com listing for Replicate

Asks the agent to install and verify. Works inside Claude Code, Cursor Agent, Codex CLI.

Install the OpenRouter MCP server for my [client: Claude] at the default config path for that client.

Use the exact install snippet published at https://top-mcps.com/mcp/openrouter (fetch https://top-mcps.com/mcp/openrouter.json for the canonical server.json if you can read URLs).

Before finishing:
1. Create the required secrets (OPENROUTER_API_KEY) and put them in the appropriate env block — do not hard-code them.
2. Restart or reload the client so it picks up the new server.
3. Verify the server is connected (green / running state) and at least one tool is listed.
4. If anything fails, read the client's MCP logs and report the exact error — do not silently retry.

Confirm when done and list the tools the server now exposes.

Frequently asked questions

What changed

— 2 updates tracked.

May 26, 2026
Refreshed install snippets and fact sheet; verified for 2026.
Jan 15, 2025
Initial directory listing.

More AI & Machine Learning MCPs

Other tools in the same category worth evaluating.

Hugging Face

Search, inspect, and run Hugging Face models and datasets from an agent.

huggingfacemlmodelsinference

3 minLow

Hugging Face

Official

Hugging Face — official ChatGPT Apps directory listing, verified for 2026.

hugging-facechatgpt-appofficialapps-directory

< 1 minLow

Replicate

Run any open-source model on Replicate from inside an AI agent.

aimlreplicateinference

3 minLow

Compared with OpenRouter

Side-by-side breakdowns for the choices people most often weigh against this MCP.

OpenRoutervsHugging Face OpenRoutervsReplicate

Exploring Top MCPs for AI & Machine Learning? See all AI & Machine Learning MCPs →

Quick answer

What it does

Best for

Not for

Setup recipe

Install

Set required secrets

Try a minimum working prompt

Tools & permissions

Security & scope

Agent prompt pack

Full details

Frequently asked questions

Is OpenRouter more expensive than going direct?

Does it support Anthropic prompt caching?

Can I see per-model cost?

What changed

More AI & Machine Learning MCPs

Hugging Face

Hugging Face

Replicate

Compared with OpenRouter