https://brtkwr.com/tags/agents/Recent content in Agents on brtkwr.comHugoenWed, 29 Apr 2026 10:30:00 +0000https://brtkwr.com/posts/2026-04-29-openclaw-heartbeat-session-pinned-model/Wed, 29 Apr 2026 10:30:00 +0000https://brtkwr.com/posts/2026-04-29-openclaw-heartbeat-session-pinned-model/<h2 id="tldr"> TL;DR <a class="heading-link" href="#tldr"> <i class="fa-solid fa-link" aria-hidden="true" title="Link to heading"></i> <span class="sr-only">Link to heading</span> </a> </h2> <p>My <a href="https://github.com/openclaw/openclaw" class="external-link" target="_blank" rel="noopener">OpenClaw</a> gateway kept calling an LLM provider I had retired. The new primary in <code>openclaw.json</code> was ignored because the long-running &ldquo;heartbeat&rdquo; session pinned the model selection at session creation time, both in the session index and in a <code>model_change</code> event at the top of the session transcript. Resetting the session let the global config take effect.</p> <h2 id="motivation"> Motivation <a class="heading-link" href="#motivation"> <i class="fa-solid fa-link" aria-hidden="true" title="Link to heading"></i> <span class="sr-only">Link to heading</span> </a> </h2> <p>I noticed my self-hosted OpenClaw gateway was still hitting an LLM provider I had switched away from days earlier. <code>openclaw.json</code> had a different primary model and no fallbacks, but the metered usage disagreed. Each &ldquo;heartbeat&rdquo; tick cost close to ten cents and shipped 370k cached input tokens to the wrong provider.</p>