OpenSquilla Open-Source Project Uses Smart Routing and Local Retrieval to Cut LLM Costs
开源项目OpenSquilla:智能路由与本地检索,大幅降低LLM使用成本
OpenSquilla combines local model routing, vector retrieval, incremental sending, and cache hits to reduce transmitted tokens by more than 90%, while routing simple tasks to cheaper models and complex tasks to stronger models without spending tokens on the routing decision.
Why it matters: OpenSquilla runs model routing decisions locally, judging question difficulty without spending tokens, and cuts actual transfer volume below a tenth through incremental sending and cache hits.