14 questions answered. Every answer sourced.
Ollama: Run Open AI Models Locally and in the Cloud
Ollama is the most popular way to run open AI models on your own computer. In 2026 it also runs larger models in its own cloud and can power coding agents like Claude Code and Codex. Here is how to install it, choose models, and use it well.
GPTPrompts.AI Editorial
Facts checked against official announcements and documentation, October 2026 · Last updated October 2026
Affiliate link, we may earn a commission at no extra cost to you.
01 · Basics
What Ollama is
A model runner, a model library and a local API in one install.
What is Ollama?
Ollama is an application and command-line tool that downloads, runs and manages open-weight large language models. It packages each model with its settings so that a single command such as 'ollama run gemma4' downloads and starts a chat. It also runs a local server so other apps can send it requests.
Source: Ollama; Ollama: Download
Is Ollama free?
Running models locally is free and Ollama is open source. Ollama's cloud models are a separate service: you sign in, and usage beyond what is included needs credits or a paid plan. In August 2026 Ollama moved its Pro, Max and Team plans to per-token pricing with usage included on every plan, so check ollama.com for current rates before relying on cloud models.
Source: Ollama Blog: Ollama's transparent pricing (Aug 31, 2026); Ollama: Download
Which models can Ollama run?
Ollama's library covers most popular open models, including Google's Gemma 4, OpenAI's gpt-oss, Qwen, NVIDIA Nemotron and Meta's Muse Glimmer 30B, plus vision, embedding and, since October 2026, speech models. Browse the library on ollama.com to see sizes and which ones are available as :cloud versions.
Source: Ollama Blog; Ollama Blog: Muse Glimmer from Meta Superintelligence Labs (Aug 10, 2026)
02 · Install
How to install Ollama and run your first model
One install command, one run command.
How do I install Ollama?
On macOS or Linux run: curl -fsSL https://ollama.com/install.sh | sh. On Windows, run in PowerShell: irm https://ollama.com/install.ps1 | iex. You can also download the desktop app from ollama.com/download. The app gives you a chat window; the command line gives you full control.
Source: Ollama: Download
How do I run a model?
Open a terminal and run ollama run followed by a model name, for example ollama run gemma4. The first run downloads the model; afterwards it starts in seconds and works offline. Use ollama list to see installed models and ollama rm to delete one you no longer need.
Source: Ollama: Download; Ollama Docs: FAQ
How do I use Ollama's cloud models?
Run ollama signin once, then run a model whose name ends in :cloud, for example ollama run gpt-oss:120b-cloud. Cloud models run on Ollama's servers, so they are fast on any computer and need no download, but every request needs an internet connection. Ollama says its cloud does not retain your data.
What hardware do I need?
Small models run on an ordinary modern laptop. Larger models need much more memory: an Apple Silicon Mac with plenty of unified memory or a PC with a high-memory NVIDIA or AMD GPU. If a model is slow or will not load, choose a smaller size or use the :cloud version. Ollama's GPU documentation lists supported cards.
Source: Ollama: Download; Ollama Docs: FAQ
03 · Build with it
Using Ollama with apps and coding agents
The local API is what makes Ollama more than a chat app.
Does Ollama have an API?
Yes. Ollama serves a REST API on your machine (by default at localhost port 11434) and offers official Python and JavaScript libraries. It is compatible with the OpenAI API and, since January 2026, the Anthropic Messages API, so tools written for those can point at Ollama by changing the base URL.
Can I use Ollama with Claude Code or Codex?
Yes. The ollama launch command, added in January 2026, sets up and runs coding tools such as Claude Code, OpenCode and Codex with local or cloud models without manual environment variables or config files. You can also configure Claude Desktop to use Ollama as a gateway provider.
Source: Ollama Blog: ollama launch; Ollama Blog
Can I use Ollama in VS Code?
Yes. Editor extensions that support local or OpenAI-compatible models can connect to Ollama's local server, giving you code completion and chat that never leaves your machine when you use local models.
Source: Ollama Docs: FAQ
04 · Compare
Ollama vs LM Studio and cloud chatbots
Pick based on how you like to work.
Ollama or LM Studio?
Both run local models for free. Ollama is command-line first with a simple app and is the usual choice for developers, scripts and coding agents. LM Studio is a polished desktop app for browsing and chatting with models through a graphical interface. Many people install both.
Is Ollama as good as ChatGPT?
The software is only as capable as the model you run. Large frontier models in ChatGPT, Claude or Gemini still lead on the hardest reasoning and long agentic tasks. Open models in Ollama are very capable for drafting, summarising, coding help and private document work, especially on strong hardware or through :cloud models.
Source: Ollama
05 · Privacy
Privacy and safety
Local by default, cloud by choice.
Is Ollama private?
Local models run entirely on your computer, so prompts and files never leave it. Cloud models send requests to Ollama's servers; Ollama states its cloud does not retain your data. If privacy is the goal, use local models only and keep cloud features off.
Is it safe to expose Ollama to my network?
By default Ollama listens only on your own machine. If you change it to listen on your network, anyone who can reach it can use your models and hardware, so restrict access with a firewall or put it behind authentication.
Source: Ollama Docs: FAQ
Sources
Read and checked October 2026. Policies change, so confirm anything time-sensitive on the source page.