Manifest
Route each query to the cheapest model on your machine
Manifest routes your AI requests to the most cost-effective model while keeping scoring on your machine. It intercepts each query, evaluates it locally in under 2 ms, and sends it to the right model to reduce costs and avoid overusing large models. You can track spending with a dashboard for costs, tokens, and alerts. Manifest exports telemetry via OpenTelemetry, offers a native OpenClaw plugin for quick installation, and is fully open source, allowing you to inspect, extend, or self-host.
