MCPs for LLM providers, embeddings, and vector stores
LLM integration MCP servers expose alternative LLM providers, embedding endpoints, and vector databases. They let one agent use GPT-4 for coding, Claude for writing, and Perplexity for research — routing per-query based on strengths. Foundation of multi-model agent systems.
LLM integration MCPs connect agents to alternative LLM providers (OpenAI, Anthropic, Mistral, Perplexity), embedding models, and vector databases. They let you route queries across models and build multi-model architectures.
Paste a research topic in Notion and an agent uses Perplexity to gather sources, summarize findings, and structure them.
Run Tavily searches on scheduled topics and index the results in Supabase for trend analysis and content research.
Run daily Perplexity searches on competitors and log product updates, pricing changes, and news to a Notion tracker.
Tool-routing: your primary agent (Claude) can call a secondary agent (GPT-4) for specific tasks. This enables cross-model workflows and specialization.
Yes — Ollama and LM Studio have MCPs that expose local Llama, Mistral, or Qwen models to Claude Code. Useful for offline or privacy-sensitive scenarios.
Yes — OpenAI and Voyage have embedding MCPs. Pair with a vector DB MCP (Pinecone, pgvector) for a complete RAG pipeline.
Each sub-LLM call adds its own token cost. For cost-sensitive workflows, route only specialized queries to premium models and keep routine calls on cheaper models.
Browse the full marketplace or explore all tags to find the right MCPs for your stack.