OrcaRouter
Let your coding agent browse 160+ LLM models and live pricing without an API key, then route chat completions through OrcaRouter when you add a key for cost-aware model picks.
Overview
OrcaRouter is a Build-phase MCP server that lets agents list 160+ LLMs with live pricing without a key and route chat via OrcaRouter when ORCAROUTER_API_KEY is set.
What is this MCP server?
- Catalog tools work with no API key: orcarouter_models_list, orcarouter_model_card, orcarouter_providers_list
- 160+ LLM models with live pricing for comparison inside the agent
- orcarouter_chat requires ORCAROUTER_API_KEY for routed completions via api.orcarouter.ai
- Configurable ORCAROUTER_BASE_URL and 300s default request timeout for long reasoning runs
- npm stdio package @orcarouter/mcp v1.1.5
- 160+ LLM models in catalog
- Default gateway: https://api.orcarouter.ai
- Default request timeout: 300 seconds
Community signal: 6 GitHub stars.
What problem does it solve?
You waste agent turns guessing which model is cheapest or capable enough because pricing and model metadata live scattered across provider sites and ad-hoc notes.
Who is it for?
Indie builders running multi-model agent workflows who want in-editor model discovery and optional centralized routing without wiring each provider separately for catalog lookups.
Skip if: Teams that have already locked a single model in CI with fixed infra and never change inference inside the IDE.
What do I get? / Deliverables
Your agent lists models and providers from one MCP surface and can send chat through OrcaRouter’s gateway after you add an API key.
- In-agent model and provider listings with live pricing metadata
- Optional routed chat completions through OrcaRouter when keyed
- Documented timeout and base URL settings for long-running agent calls
Recommended MCP Servers
Journey fit
Model catalog and routing are core agent-tooling decisions made while you wire integrations and choose inference backends during Build. Agent-tooling is the right shelf because OrcaRouter is about which model to call and at what price—not shipping UI or running production analytics.
How it compares
LLM catalog and routing MCP integration, not a prompt library skill or local inference runtime.
Common Questions / FAQ
Who is OrcaRouter MCP for?
Solo builders and small teams using Claude Code or Cursor who want the agent to compare 160+ models and live pricing, optionally routing chat through OrcaRouter.
When should I use OrcaRouter MCP?
Use it while building agent tooling—choosing models for codegen, planning, or review—and whenever you need up-to-date pricing before committing to a provider.
How do I add OrcaRouter to my agent?
Add stdio MCP server @orcarouter/mcp from npm in your client config; set ORCAROUTER_API_KEY only if you need orcarouter_chat. Optional: ORCAROUTER_BASE_URL and ORCAROUTER_REQUEST_TIMEOUT.