Ollama is an open-source software platform for running and managing large language models (LLMs) on local computers and through hosted cloud models. It was developed by Jeffrey Morgan and Michael Chiang and first released on July 7, 2023.
Overview
Ollama provides a command-line interface, a native GUI, a local REST API, and model-management tools that allow users to download, run, import, and manage open-weight large language models. It is often described as "Docker for LLMs" because it simplifies the process of pulling and running models with minimal configuration.
Key Details
| Attribute | Value |
|---|---|
| Developers | Ollama Inc. and contributors |
| Initial release | July 7, 2023 |
| Stable release | 0.22.1 (April 28, 2026) |
| Written in | Go |
| Operating system | macOS, Linux, Windows |
| License | MIT License |
| Website | ollama.com |
| Repository | github.com/ollama/ollama |
History
Ollama was first released in 2023 and quickly became associated with the growth of local LLM software. It supports models such as Llama, Gemma, Mistral, Qwen, DeepSeek, GLM, and others. In 2025 and 2026, Ollama added hosted cloud models, web search support, tool and coding-agent integrations, and compatibility with applications such as Claude Code, Codex, OpenCode, and Copilot CLI. In March 2026, it announced preview support for Apple silicon. In July 2026, the company raised $65 million in funding and reported nearly 9 million users.
Features
- Local model inference using the llama.cpp backend, supporting quantized models that use less memory.
- GPU acceleration via CUDA (NVIDIA), ROCm (AMD), and Metal (Apple).
- REST API exposed by default on port 11434 for chat and model management.
- Client libraries for Python and JavaScript.
- Official Docker image for containerized deployments.
- Model library allowing users to pull, run, and manage model variants by name.
- Experimental image generation support using models such as Flux (currently on macOS).
Security
Because Ollama is designed to run locally, security researchers have noted risks from misconfigured public deployments. In January 2026, SentinelOne and Censys reported that many Ollama servers were exposed to the public internet by binding to 0.0.0.0 without adequate security measures.
