OpenClaw Ate Your Lunch
OpenClaw dominates OpenRouter's Claude 3.5 Sonnet token chart, but total demand hasn't grown; it has been displaced. The early-adopter wave may already be cresting.
I was poking around the OpenRouter dashboards today and noticed something worth a closer look.
Pull up the Claude 3.5 Sonnet activity chart. The total daily token burn has stayed flat over the past month, but the composition has completely changed. OpenClaw (the blue band) has swollen to dominate the chart. 14.4 billion tokens this month. The next app, Cline, sits at 1.8 billion.
At first glance it looks like massive growth. But zoom out and the total bar height has barely moved. OpenClaw hasn't grown demand for Claude 3.5 Sonnet; it has displaced it. All those tokens flowing through Cline, Roo Code, liteLLM, and others? They're now flowing through OpenClaw instead.
Which raises a question: what were we all doing before that we're now doing through OpenClaw?
Drill into OpenClaw's own page on OpenRouter and the growth curve tells the same story. The hockey stick from late January is bending into a plateau. The early-adopter wave is cresting.
Between the security concerns, the founder joining OpenAI, and users figuring out that running an always-on agent through a $15/million-token model adds up fast, it feels like we're entering the "now what?" phase of agentic AI's first mainstream moment.
Maybe that's exactly where we need to be. The hype wave showed us what autonomous agents can do. The plateau is where we figure out how to make it sustainable.
What are you seeing in your own usage patterns? Has OpenClaw changed how you work, or was it more of a "try it and move on" moment?
See for yourself:
Data: OpenRouter public dashboards, 23 Feb 2026