Mesh Playground

Run inference on the live mesh.

Send a prompt and watch it route to a real node, execute, and return — with full routing metadata, cost savings, and telemetry.

Syncing ecosystem state…

—

Prompt

Model

Cost preference

6 ms local fast path

Set privacy_policy: "local_only" to hit the 6 ms benchmark route — no auth, no node lookup, no LLM call. The response is generated on-device instantly with fire-and-forget telemetry logging.

Build against this endpoint

Call base44.functions.invoke('meshInference', { prompt, model, cost_preference, privacy_policy }) from any app to run inference on the live mesh. Every call routes to a real node, updates its task counters, and logs routing + telemetry.