ollama
Run open AI models locally or in the cloud, chat with them, and connect them to coding tools and business apps.
- Stars
- 186k
- Forks
- 19k
- Last update
- 5 Nov 2026
Overview
Ollama runs open models such as Gemma, Qwen and DeepSeek and provides a REST API for running and managing them. You can chat from the command line or connect models to coding assistants, messaging assistants and community-built interfaces.
Who it's for
Tech stack
Features and benefits
Key features
- Local and cloud access to open models
- Command-line model chat and interactive setup
- REST API for running and managing models
- Official Python and JavaScript libraries
- Claude Code connection through an Anthropic-compatible API
- Launch integrations including Codex, Copilot CLI and OpenCode
- OpenClaw integration for assistants across WhatsApp, Telegram, Slack and Discord
- Installers for macOS, Windows and Linux, plus an official Docker image
- Community integrations for chat interfaces, document retrieval and development tools
Why it helps in Pakistan
- Choose local model execution when you want to keep processing on your own machine.
- Add model chat to existing applications using the REST API or official libraries.
- Use open models with familiar coding assistants.
- Choose cloud models when you want larger models without downloading them.
- Build on existing community interfaces rather than creating every component yourself.
Pakistan use cases
- 01
A Lahore software house can connect Claude Code to Ollama for repository questions and code edits.
- 02
A Karachi developer can add model chat to an internal business app using the Python or JavaScript library.
- 03
An Islamabad consultancy can explore PDF question-answering through the listed Ollama RAG Chatbot integration.
- 04
A Pakistani small-business owner can use OpenClaw with Ollama as a personal assistant across WhatsApp and Telegram.
Requirements
- Hardware
- Check the official docs for RAM, storage and GPU requirements; the supplied sources do not give minimum specifications.
- Operating system
- macOSWindowsLinux
- Skills
- Basic terminal use for the documented commandsPython, JavaScript or REST API knowledge for application integrationLinux service administration for a systemd deployment
- Accounts needed
- Check the official docs for account requirements.Direct Claude Code access to Ollama cloud requires an Ollama API key.
Setup guide
Commands are copied from the project's README and official docs.
- 1
Install Ollama
On macOS, runcurl -fsSL https://ollama.com/install.sh | sh. On Windows, runirm https://ollama.com/install.ps1 | iex. The README also links manual installers. - 2
Open the interactive launcher
Runollamaand follow the prompt to run a model or connect an existing application. - 3
Chat with a model
Runollama run gemma4to chat with Gemma 4. - 4
Optionally connect a coding assistant
Runollama launch claudeto launch the Claude Code integration.
Works with
Connect local or cloud models through Ollama's Anthropic-compatible API using `ollama launch claude`.
Run `ollama launch openclaw` to use Ollama with a personal assistant across WhatsApp, Telegram, Slack and Discord.
Install with `pip install ollama` and use the documented chat function.
Install with `npm i ollama` and call ollama.chat from JavaScript.
Listed as a community integration providing an extensible, self-hosted AI interface.
Listed as a community integration for chatting with multiple PDFs using RAG.
What it costs
| Software | Free to self-host under the MIT licence. |
|---|---|
| Hosting | Check the official docs before budgeting for hosting. The supplied sources do not state hardware requirements, so a reliable VPS estimate cannot be sized from them. |
| Paid cloud option | Official Ollama cloud models are mentioned. Check the official service for current terms and pricing; no prices are supplied. |
FAQ
Can I run models locally?
Yes. Ollama supports local model execution as well as cloud models. The README demonstrates local chat with ollama run gemma4.
Does Ollama include a WhatsApp business inbox?
The sources describe WhatsApp access through the OpenClaw personal-assistant integration, not a built-in business inbox.
Can I use it in a Python or JavaScript application?
Yes. Official libraries are available for both languages, along with a REST API.
How much RAM or GPU memory do I need?
Check the official docs. The supplied sources do not specify minimum RAM or GPU memory requirements.
What context setting is recommended for Claude Code?
The Claude Code guide recommends a 64k or higher context window for local models and larger repositories.
Troubleshooting
The Linux service is not running.
Run sudo systemctl start ollama, then sudo systemctl status ollama. Inspect service logs with journalctl -e -u ollama.
Claude Code connects to the wrong endpoint or uses old credentials.
Claude settings can override shell variables. Update the base URL and credentials in those settings, then use /status to check the connection.
AMD GPU support is incomplete on Linux.
The Linux guide recommends ROCm v7 and the latest AMD Linux driver. Manual AMD installations also need the additional ROCm package described in the official docs.
Ollama needs updating on Linux.
The documented update method is to rerun curl -fsSL https://ollama.com/install.sh | sh.
Official links
Get it set up for you
Don't want to install it yourself? A Jugaad maker can set up ollama for your business. Payment is held in Safe Deal until it works.
No maker offers a setup gig for this yet. Post a request and makers will quote.
Related tools
Built by ollama and contributors on GitHub. All credit goes to the authors. GitHub stats refresh daily. Jugaad does not host, copy or sell this code.
Is this your project?