A pre-launch Geekbench 7 result points to the hardware configuration behind OpenAI's dots autonomous agents: a single nine-core slice of an AMD EPYC 9V74 with 9.73GB of memory. The run, posted on X by user INIYSA and captioned "OpenAI dots Geekbench 7 Score," shows a single-core score of 1,667 and a multi-core score of 9,435. The result was uploaded Sept. 25 — four days before the product's launch. The original tester's identity remains unknown.

A check of Geekbench 7's public database found six runs on that same configuration uploaded since launch, ranging from 1,512 to 1,614 single-core and 8,135 to 8,991 multi-core. The benchmark ran on Ubuntu for the leaked run and Debian for the post-launch runs. The pre-launch 9,435 multi-core score sits almost 5% above the post-launch best and roughly 10% above the median, making it a potential outlier, though performance remains in the same ballpark.

Dots, powered by GPT-6 Astra, launched at OpenAI's DevDay on Tuesday for Pro and Business Premium users. OpenAI bills the agents as having "their own cloud computer, their own browser, and the apps you've connected." The concept 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 a list of apps her dot reported as preloaded: Chromium, Blender, GIMP, Inkscape, Kdenlive, Godot, FreeCAD, OpenSCAD, KiCad, QGIS, ParaView, and 3D Slicer, alongside Python, Node.js, and Git. In the same reply, the dot said its cloud computer runs "Debian Linux."

The performance gap against Meta's rival Muse is substantial. Tom's Hardware previously reported that Muse runs on AMD EPYC Turin with two cores and 8GB per sandbox, reporting 7.75GB of memory in its Geekbench runs. 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 single-core and about 8,550 multi-core, land at roughly 1.5x Muse's single-core and 6x its multi-core result.

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

Data center CPU demand for always-on agents has surged, as high-core-count CPUs are needed at sustained load. It is unknown whether dots, as tested, hold their cores while idle. Adoption is accelerating: Muse reportedly passed 500,000 daily active users, and SpaceXAI plans standalone Nvidia Vera CPUs for Grok's agentic workloads. A new CPU arms race may only be a matter of time. OpenAI has already said users will eventually be able to add more dots and "scale the output of each dot by ... increasing its speed."