Ollama

by Ollama

Run open-source LLMs locally with a simple developer-first platform

See https://ollama.ai

Features

  • Local-first model serving: download and run open-source models (Llama 3, Gemma, Mistral, Phi) locally
  • Command-line interface and REST API for programmatic access
  • Function calling / tool calling support for models
  • Structured output (JSON schema enforcement) for reliable programmatic parsing
  • Multimodal support (text + images) and expanded image format compatibility
  • Model management: install, list, remove, and configure models; mod files for custom configs
  • OpenAI-compatible API layer for easier migration and integration
  • New standalone desktop app for non-technical users (model browsing, chat, file-based chat)
  • MCP server ecosystem support for using Ollama models in MCP-based agent frameworks

Superpowers

  • Privacy & offline operation: run LLMs without sending data to the cloud
  • Developer ergonomics: simple CLI + mod files make local deployment fast and scriptable
  • Production readiness: health checks, logging, and performance optimizations for on-prem use

Pricing

  • Free open-source tooling

Quick usage examples

  • Run a local model via CLI and query it from a REST client
  • Use function calling to integrate local models with tools and databases
  • Deploy an MCP server to let agent frameworks use local Ollama models

Sources

  • Ollama official website and changelogs
  • Community tutorials and MCP implementations