# Outrun blog

Guides and explainers on AI models, coding agents and spending less on both.

- [Why use Zo Computer, and how to set it up](https://outrun.to/blog/zo-computer): Zo Computer is your own always-on computer in the cloud with an AI agent built in. Here is what it can do, what it costs and how to set it up.
- [Why coding agents refuse when abliterated models don't](https://outrun.to/blog/why-harnesses-refuse-abliterated-models): You point a coding agent at an abliterated model and it still says no. The refusal usually comes from the harness around the model, not the model itself.
- [Why long agent sessions cost more, and how to cut the bill](https://outrun.to/blog/save-money-on-ai-agent-context): Your coding agent rereads the whole conversation on every turn. Here is how caching helps, what breaks it, and how working in stages cuts the bill.
- [How do I point Claude Code at a custom base URL?](https://outrun.to/blog/claude-code-custom-base-url): Claude Code reads its API address from one setting. Change two values and it works with any Anthropic-compatible API, Outrun included.
- [Uncensored vs abliterated models: what is the difference?](https://outrun.to/blog/uncensored-vs-abliterated-models): Both kinds answer requests that mainstream chat models turn down. They get there in different ways, and that changes which one to use.
- [How do I set up my own AI image and video studio?](https://outrun.to/blog/local-image-video-studio): A studio on your own computer gives you every setting each model supports and keeps your key local. One setup prompt builds it.