Ollama

Ollama bundles model weights, configurations, and runtime dependencies into a streamlined CLI and local server for running open-source LLMs locally with native tool-calling.

When to Use

• Local LLM Execution: Running Llama 3, DeepSeek-R1, Mistral, and Qwen locally with a single CLI command.
• Zero-Cloud Air-Gapped Deployments: Privacy-critical enterprise environments requiring 100% on-premise execution.
• Custom Modelfile Configuration: Tailoring system prompts, temperature, context length, and stop sequences.
• REST & OpenAI-Compatible API: Seamless drop-in replacement for OpenAI SDKs in local test and dev pipelines.

Quick Start

# Run open-weight models locally with zero setup
ollama run llama3.2:3b "Write a Python script to compute Fibonacci numbers."