When you set up Skales, the first decision is which AI provider to connect. The two most common choices are OpenRouter (cloud-based, wide model selection, pay-as-you-go) and Ollama (local, free, fully offline). Both integrate natively with Skales. Here is how to choose.
When to Use OpenRouter
OpenRouter is the easier starting point. Create a free account, get an API key, paste it into Skales — you are up in two minutes. OpenRouter gives you access to GPT-4o, Claude 3.5 Sonnet, Llama models, Mistral, Gemini, and many others through a single API key. You pay per token consumed, which for most users is $3-10 per month. Choose OpenRouter when: you want the best possible model quality, you have a low-powered machine that struggles with local models, you want easy access to multiple frontier models, or you do not process sensitive personal data.
When to Use Ollama
Ollama runs the model entirely on your hardware. No API costs, no data leaving your machine, works without internet. Setup takes about 10 minutes (install Ollama, pull a model, connect it in Skales Settings). Choose Ollama when: you process sensitive or confidential data, you want guaranteed offline operation, you have a machine with 16GB+ RAM and want fast local responses, or you want $0 AI costs ongoing. The models available via Ollama (Llama 3.1, Mistral, Qwen 2.5, DeepSeek) are excellent for everyday tasks.
The best setup for most power users: Ollama as the default provider for sensitive and daily tasks, OpenRouter as a secondary provider for occasional queries needing frontier capability. Skales supports both simultaneously. Download free and connect both in under 15 minutes.