How I'm Solving Local Inference

This article details the author's journey to solve local inference for AI models by connecting two laptops—a MacBook Air and a Framework 13—using LM Studio's LM Link feature. The author discusses the rising costs of per-token billing from frontier AI labs like OpenAI and Anthropic, and the improving quality of local models from Moonshot AI, DeepSeek, and Alibaba. Facing hardware constraints with 24 GB RAM on the MacBook Air, they leverage the Framework 13's 64 GB RAM to run models like qwen3-coder-next remotely via LM Link. The article covers setting up LM Studio, using the lms CLI tool, and replacing cloud-based tools like Claude Code with a local inference setup. It's a practical guide for developers interested in running AI models locally to avoid variable token fees.