Runs open-weight language models locally behind a local API
Ollama runs open-weight language models on your own machine with a single command, then serves them over a local HTTP API. Its library covers Llama, Mistral, Gemma and Qwen variants plus embedding and vision models, and a Modelfile format pins system prompts and sampling settings into a custom variant. Because it exposes an OpenAI-compatible endpoint, existing code and desktop clients can point at localhost instead of a paid service. The trade-off is expected: quality and speed depend on the memory your computer can spare.
Last updated: 2026-09-20. This site only provides an index; for exact features, pricing, and licensing, see the official website.
local llm inference, private assistants, offline prototyping and self-hosted model serving
If you're comparing similar products, check the alternatives below, or browse all tools in the AI Models & Platforms category.
Its library covers many open-weight families, including Llama, Mistral, Gemma and Qwen variants, plus embedding and vision models. Choice depends on memory: small quantised models run on laptops, while larger ones need a workstation with serious video memory.
Running models locally costs nothing beyond your hardware and electricity. There is also an optional hosted plan for larger models or machines that cannot handle them, billed separately, so check that page if you need cloud capacity.
Yes. Ollama serves an HTTP API on your machine, including an OpenAI-compatible endpoint, so many clients and libraries can be pointed at localhost instead of a paid provider with only a base URL change.
Developer console and API keys for the Claude model family
Node-based local interface for running image and video diffusion models
Browser playground for prompting and prototyping with Gemini models
Hosted Jupyter notebooks with optional free GPU and TPU runtimes