FIELD NOTEUpdated from community signals on 2026-03-18.

Field Signal

Most OpenClaw users burn money without realizing why -> heartbeat and fallback loops are the real cause.

Why OpenClaw cost gets so high (and how to fix it)

Quick Read

Most setups do not fail because of bad models.

They fail because of silent token drain.

Heartbeat. Fallback. Agent loops.

What's causing it

Heartbeat frequency

Intervals are often too aggressive, so token usage keeps running while you are idle.

Fallback triggers

Fallback is useful, but under load it can activate far more often than expected.

Multi-agent loops

Costs multiply quickly when more than one agent is running the same traffic pattern.

Seen repeatedly in the community

Community threads report budgets drained in minutes without clear warning.

Users repeatedly ask if their setup is viable only after the spend has already happened.

Manual provider caps are commonly used as emergency guardrails.

Real example

Heartbeat every 30 minutes -> 48 calls/day -> roughly $90/month even when idle.

Tool bridge

Use the calculator to see your real cost before running anything.

Open Calculator

Other ways to fix this

Choosing a cheaper model can reduce cost significantly.

Try Model Picker

Related Field Notes

Part of GuardClaw. Field notes and tools for OpenClaw developers.