A Geekbench 7 result uploaded on September 25 — four days before OpenAI launched its dots agents at DevDay on Tuesday — shows a multi-core score of 9,435 on what appears to be a nine-core slice of an AMD EPYC 9V74 with 9.73GB of memory. Shared on X by INIYSA, the pre-launch run also posted a single-core score of 1,667. Post-launch runs on the same configuration land well below it but stay in the same ballpark.

A check of Geekbench 7's public database found six runs on that configuration uploaded since launch, ranging from 1,512 to 1,614 single-core and 8,135 to 8,991 multi-core. The pre-launch 9,435 is almost 5% above the post-launch best and about 10% above the median. The original runner is unknown. The leak ran on Ubuntu; post-launch runs used Debian, which a dot itself confirmed runs its cloud computer.

What hardware powers each dot?

The VM appears to be a single nine-core slice of an AMD EPYC 9V74, listed by Geekbench with a 2.6GHz base clock and 9.73GB of memory. Dots, powered by GPT-6 Astra, launched at OpenAI's DevDay on Tuesday for Pro and Business Premium users.

OpenAI bills the autonomous agents as having "their own cloud computer, their own browser, and the apps you've connected." The idea is not new. Cursor says its cloud agents get their own virtual machines; SpaceXAI says every Grok Bot shares one persistent cloud computer; and Meta's Muse gives each agent a sandbox.

Maddie D. Reese posted on X on launch day the apps her dot said came preloaded. The lineup includes:

• Chromium, Blender, GIMP, Inkscape, Kdenlive
• Godot, FreeCAD, OpenSCAD, KiCAD
• QGIS, ParaView, 3D Slicer
• Python, Node.js, and Git

How big is the gap versus Meta's Muse?

Tom's Hardware previously reported that Meta's Muse runs on AMD EPYC Turin with two cores and 8GB per sandbox. The 10 Muse runs in Geekbench's database show median scores of about 1,041 single-core and 1,394 multi-core. The dot's post-launch runs, with medians of 1,570 and about 8,550, put it at roughly 1.5x single-core and 6x multi-core.

Some of that gap comes from the nine-core versus two-core configuration. The rest likely reflects clock speed: Geekbench lists 2.6GHz base for the dot's EPYC 9V74 against 1.5GHz for Muse's EPYC 9D25, partially offset by Muse's newer architecture. Muse's runs report 7.75GB of memory. An independent run for this article returned 1,000 single-core and 1,433 multi-core, in line with the other Muse results.

Why does this matter for CPU demand?

Always-on agents are reshaping data center CPU requirements because sustained load favors high-core-count processors. It is unknown whether dots, as tested, hold their cores while idle. The agent category is scaling fast: Muse has reportedly passed 500,000 daily active users, and SpaceXAI plans standalone Nvidia Vera CPUs for Grok's agentic workloads.

OpenAI has already signaled where this goes next, saying users will be able to add more dots and "scale the output of each dot by ... increasing its speed." If agent fleets keep growing, a new CPU arms race over cores-per-agent and sustained-clock server silicon may only be a matter of time.