The creator of OpenClaw used $1,300,000+ of OpenAI tokens in 30 days, which is a hell of a perk
OpenClaw creator Peter Steinberger incurred a $1,305,088.81 bill for OpenAI API tokens in just 30 days, according to a screenshot of his OpenAI dashboard he posted publicly. The massive expenditure represents 603 billion tokens consumed across 7.6 million API requests, with GPT-5.5 being the primary model used. The infrastructure supporting this volume—approximately 100 Codex instances—was managed by a team of just three people.
This staggering figure illustrates the computational cost of operating large-scale AI agent systems. For context, processing millions of requests in a single month requires substantial API spending, and Steinberger’s usage demonstrates the scale at which AI models can consume tokens when deployed for continuous operations. The data provides a rare public glimpse into the actual operating costs of an AI-powered application, showing that running infrastructure dependent on OpenAI’s models at significant scale can easily translate into seven-figure monthly expenses.
While the exact purpose and nature of OpenClaw’s operations aren’t detailed in the available excerpt, the usage statistics highlight broader conversations about AI infrastructure costs and sustainability in production environments.