TL;DR
OpenClaw creator Peter Steinberger spent $1.3 million on OpenAI API tokens in 30 days running 100 Codex instances on his open source project. The bill, covered by OpenAI, where Steinberger now works, represents 603 billion tokens across 7.6 million requests and provides the most concrete public data on the cost of autonomous AI coding at scale.
Peter Steinberger, creator of OpenClaw and OpenAI engineer, racked up $1.3 million in API costs in a single month by running approximately 100 Codex instances simultaneously on his open source project. The bill, which covered 603 billion tokens across 7.6 million requests over 30 days, is the most visible demonstration yet of what happens when AI-powered software development is run without budget constraints, and how quickly costs rise when autonomous agents continually operate at scale.
Steinberg posted a screenshot of the bill on Xshowing $1,305,088.81 uploaded to the OpenAI API, with GPT-5.5 as the primary model. OpenAI covers the cost: Steinberger joined the company in February 2026, and the expense is treated as an investment in research to understand what software development is like when the token economy is not a limiting factor.
Peter Steinberger
What Agents Really Do
The 100 Codex instances are not limited to generating code. Steinberger’s three-person team has built an autonomous development process in which AI agents perform a variety of tasks that would normally require a much larger engineering organization. Agents review pull requests, scan commits for security vulnerabilities, deduplicate issues from GitHub, write fixes, and open new pull requests based on the broader project roadmap. Others monitor performance benchmarks and flag regressions on the team’s Discord server. Some agents, according to The Decoder, even attend meetings and generate feature pull requests that come up in the conversation.
The 💜 of EU technology
The latest rumors from the EU tech scene, a story from our wise founder Boris and some questionable AI art. It’s free, every week, in your inbox. Register now!
The team also uses Clawpatch.ai, Vercel’s Deepsec, and Codex Security for additional bug and security analysis. The result is a development operation in which three humans oversee a fleet of AI agents that collectively perform the work of what would traditionally be a mid-sized engineering team.
The question of cost
Steinberger has been transparent about the economics. He clarified that the figure of 1.3 million dollars reflects the Codex decision of “Quick mode“, which consumes credits at a significantly higher rate than standard execution. Disabling Fast Mode alone would reduce API’s gross cost to approximately $300,000 per month, a 70 percent reduction. At standard pricing, the operation would still cost $3.6 million a year, but the gap between the headline figure and the underlying economics illustrates how pricing tiers and execution modes can dramatically inflate reported costs.
When asked about return on investment, Steinberger said everything his team builds is open source and works with leading proprietary models as well as open-source alternatives. “I would say pretty high“, said.
The figure is useful precisely because vendor marketing around AI coding tools rarely reveals gross spending and token volumes on this scale. Most enterprise teams planning agent development tools work from projections and estimates. Steinberger’s bill is public and hard data: 100 agents running continuously for 30 days on a large open source database costs between $300,000 and $1.3 million a month, depending on execution speed, before any optimization.
Who is Peter Steinberger?
Steinberger is no newcomer to building development tools at scale. The Austrian engineer founded PSPDFKit in 2011, pioneering a PDF rendering and annotation framework that became the standard for mobile document handling. In 2021, apps built on PSPDFKit were running on more than 1 billion devices worldwide and the company raised $116 million from Insight Partners, its first outside investment after a decade of profitable, self-funded growth.
After leaving PSPDFKit, Steinberger began experimenting with AI agents as a personal project. OpenClaw, a self-driving AI assistant that runs entirely on users’ own hardware, became the Fastest growing open source project in GitHub historysurpassing 302,000 stars in April 2026, surpassing React, Vue.js, and TensorFlow in a fraction of the time it took those projects to reach similar milestones. The framework connects to tools people already use, including email, calendars, browsers, and messaging platforms, from Slack and Discord to WhatsApp and iMessage, and allows AI agents to execute shell commands, manage files, and automate web tasks locally.
When Steinberger joined OpenAI, he announced that OpenClaw would be moved to an independent foundation to preserve its open source character. “I want to change the world, not build a big company,“, wrote.”Teaming up with OpenAI is the fastest way to bring this to everyone.”
What it reveals about the economics of AI coding
The $1.3 million bill comes at a time when the economics of AI-driven development is a central concern of the software industry. OpenAI recently opened ChatGPT subscriptions to OpenClaw’s 3.2 million usersallowing them to run autonomous agents through the Codex endpoint for $23 per month. Anthropic, by contrast, prevented Claude Pro and Max subscribers from using OpenClaw and other third-party agent frameworks, concluding that the computing demands of autonomous agents executing thousands of API calls per day were economically unsustainable with a flat-rate subscription price.
The divergence between those two approaches reflects an unresolved tension in AI pricing. Subscription models are designed for human-speed interaction: a person typing queries into a chat window generates a predictable and manageable volume of API calls. A fleet of autonomous agents generates orders of magnitude more, and the gap between the subscription price and the actual computing costs is the subsidy absorbed by the provider or paid by the user.
Steinberger’s bill makes that gap visible. At $1.3 million per 100 agents, the cost per agent is approximately $13,000 per month, far more than any subscription plan covers. Even with the $300,000 optimized, each agent costs approximately $3,000 per month. For enterprise teams evaluating whether to deploy agent coding tools at scale, these numbers provide a foundation that no vendor marketing page will offer.
The broader pattern
OpenClaw’s journey, from a personal experiment to the most prominent project on GitHub and an OpenAI-sponsored research platform, reflects a broader shift in the way software is built. AI coding agents from DeepMind, OpenAI and Anthropic are moving from proof-of-concept demos to production deployment, and the question is no longer whether AI will write significant amounts of code, but rather how much it will cost and who will pay for it.
The rise of AI-assisted developmentfrom individual coding co-pilots to fully autonomous agent fleets, it is compressing the timeline between the ambition of a three-person team and the output of a large engineering organization. Steinberger’s setup, three humans and 100 agents, is an extreme version of what many companies will try at smaller scales over the next year.
The $1.3 million bill is not a warning. It’s a receipt from the future, showing what it costs when AI development tools are used at full capacity, without the budget constraints that currently limit most teams to a fraction of what the technology can do. Whether that future is affordable depends on how quickly model inference costs decline, how efficiently agent orchestration frameworks manage token usage, and whether the Security and quality challenges of AI-generated code. can be managed at the speed that these agents produce it.
