Ollama is a tool for building with open models, supporting various integrations and a REST API. It is built in Go and licensed under MIT, with a focus on AI models like Kimi-K2.6 and GLM-5.1.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Ollama is a tool designed for building with open models, including Kimi-K2.6, GLM-5.1, MiniMax, DeepSeek, gpt-oss, Qwen, and Gemma.
How it works
Ollama allows users to run models and connect to existing agents or applications. It supports integrations with various tools such as Claude Code, OpenClaw, Codex, and Copilot. Users can launch specific integrations using commands like ollama launch claude or ollama launch openclaw.
Getting started
To install Ollama, users can execute the following commands based on their operating system:
- macOS:
curl -fsSL https://ollama.com/install.sh | sh - Windows:
irm https://ollama.com/install.ps1 | iex - Linux:
curl -fsSL https://ollama.com/install.sh | sh - Docker: The official Docker image is available at
ollama/ollama.
To start using Ollama, run:
ollama
To chat with a model, such as Gemma 4, use:
ollama run gemma4
For REST API usage, an example command is:
curl http://localhost:11434/api/chat -d '{"model": "gemma4", "messages": [{"role": "user", "content": "Why is the sky blue?"}], "stream": false}'
Recent releases
The latest releases include:
- v0.32.0 (2026-07-11): Added support for the qwen3.5 parser and renderer.
- v0.31.2 (2026-07-06): Enabled flash attention on older NVIDIA GPUs.
- v0.31.1 (2026-06-30): Improved performance for Gemma 4 on Apple Silicon.
Traction
- Stars: 175938
- Forks: 16940
- Open Issues: 3423
Behind the repo
The repository is linked to the startup Ollama, categorized under Developer Tools.
Caveats
- License: MIT
- Open Issues: 3423
- Created: 2023-06-26
- Last Push: 2026-07-10