Show HN: Smart model routing directly in Claude, Codex and Cursor

According to Hacker News, Weave has released a model router that sits between coding agents like Claude Code and Cursor and your LLM endpoints, intelligently deciding which model to use for each task. The problem: tools like Opus 4.8 are powerful but expensive, especially for routine work like gathering code context or exploring a repository. The router solves this by using cheaper models—DeepSeek, GLM, and Kimi—for simpler requests while reserving frontier models for genuinely complex planning and implementation. Weave reports a forty percent reduction in token costs after a month of internal use, with no noticeable drop in quality or velocity. The router is available open-source under Elastic License 2.0, or you can use their hosted version at weaverouter.com—a practical answer to the rising cost of continuous AI-assisted coding.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton