Who paid for the $1.3M OpenAI bill?

OpenAI covered it. Peter Steinberger joined OpenAI in February 2026 and uses OpenClaw as a stress-test lab.

Is $1.3M the real cost?

The figure uses Fast Mode pricing. Standard mode would be ~$300K for the same workload.

How many coding agents were running?

~100 Codex instances operated by a 3-person team running continuously for 30 days.

What does this mean for normal OpenClaw users?

Normal users spend $500-$5,000/month. The $1.3M figure is an extreme case with unlimited OpenAI budget.

← Back to dashboard

clawsmith.com/signal/openclaw-creator-1-3m-openai-api-burn-may-2026

🔥 HypeWide OpenLive

OpenClaw Creator Burns $1.3M in OpenAI API Tokens in One Month

Q: How much did the OpenClaw creator spend on API tokens?

$1,305,088 in one month — 603 billion tokens across 7.6 million requests from ~100 Codex instances.

Peter Steinberger posted a screenshot showing $1,305,088 in OpenAI API spending over 30 days — 603 billion tokens across 7.6 million requests from ~100 Codex instances operated by a 3-person team. OpenAI covers the cost. Clarified that Fast Mode pricing inflates the figure; standard mode would be ~$300K.

Product Idea from this Signal

A background service that enforces per-agent token budgets, kills runaway retry loops, and alerts before API bills explode

3.1k ▲

OpenClaw ships with zero spending controls. After Anthropic forced 135,000+ instances onto API billing in April 2026, users discovered agents burning $200/day from infinite retry loops and 200K+ token context replays on every request. The OpenClaw creator himself burned $1.3M in one month running 100 Codex instances. This service sits between OpenClaw and LLM providers, enforcing per-agent and per-session token budgets with automatic circuit breakers that kill runaway loops before they drain the wallet.

CLIOPEN-SOURCECOST-CONTROLDEVTOOLPROXY

CompetitiveView Opportunity →