Model Releases
RIP Anthropic API fees. Someone just made Claude Code run 100% locally on a MacBook for $0/month. 122B parameter model. 65 tokens per second…
RIP Anthropic API fees. Someone just made Claude Code run 100% locally on a MacBook for $0/month. 122B parameter model. 65 tokens per second. Nothing touches the cloud. The trick everyone else missed:
RIP Anthropic API fees. Someone just made Claude Code run 100% locally on a MacBook for $0/month. 122B parameter model. 65 tokens per second. Nothing touches the cloud. The trick everyone else missed: Every other local Claude Code setup uses a proxy to translate between Claude Code's API and local models. That proxy was the bottleneck the whole time. This repo wrote a 200-line server that speaks Anthropic API natively. No proxy. No translation layer. No 133-second wait times. Just 17.6 seconds per task on Apple Silicon. What you need: → M2/M3/M4/M5 Max with 64–128 GB unified memory → Python 3.12+ → Claude Code installed → One command to set it up Bonus: control it from your iPhone via iMessage. Your Mac sits at your desk. You send a message from the couch. Claude Code runs the task. Response comes back to your phone. Works offline. No API keys. No subscriptions. 100% Open Source. MIT License.
Source: Clem Delangue (X) | 2026-04-18