Instrument pentru rularea modelelor AI deschise local sau în cloud, pe Mac, Windows și Linux
Modele IA Modele și ponderi IA ollama.com
What is Ollama
Ollama is a tool for running open-source large language models on your own computer: one command downloads and starts a model. It supports macOS, Windows and Linux, and there is an official Docker image. Its repository lists support for Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models. The code is open source under the MIT licence and has more than 180,000 stars on GitHub, making it one of the most widely used tools for running models locally.
What it can do
- Run models locally: download and run with one command; inference happens on your machine and data stays on your computer
- A local API: once running it exposes an interface that many clients, such as Open WebUI, Chatbox and Cherry Studio, can connect to directly
- Cloud models: for models too large for your computer, Ollama's cloud service bills by token
- A large community has built integrations around it, such as search engines, coding assistants and knowledge-base Q&A
Pricing
Running models on your own hardware is always free and unlimited. Official plans for cloud models:
- Pay as you go: from US$0, priced per million tokens, no subscription
- Pro: US$20 a month (US$200 a year), including US$60 of usage a month
- Max: US$100 a month, including US$300 of usage a month
- Team: from US$500 a month with unlimited users and US$1,000 of shared usage a month
Who it's for
- Anyone who wants to use AI offline or with sensitive material, without sending data to the cloud
- Developers who want to try open models for free
- People who want a local model backend for their own apps
Good to know
- Local speed depends on your GPU and memory; large models may not run or may be slow, so start with smaller ones.
- Each model has its own licence; check it before commercial use.
Similar tools
Compare Jan and Chatbox AI; to download models, see Hugging Face. This site has a more detailed Ollama article and a guide to running large models locally.