Cua Adds a Guide for Running OSWorld Benchmarks on Its Fleets
- Source
- x.com
- Date
1/ You can now run OSWorld tasks on Cua Cloud Fleets. Our new guide shows how to turn the official OSWorld Ubuntu image into a Fleet desktop with Cua Driver, then run your agent and evaluate the tasks. Get started: https://t.co/T3Xx7ZcasW
2/ Bring the benchmark desktop with you. Start with the official OSWorld image and its applications. Add Cua Driver, publish the VM disk to a registry, and claim a desktop through the Sandbox SDK.
3/ Set up the task. Run your agent. Evaluate the result. Use OSWorld's setup tools and evaluators against the Fleet desktop, with Cua Driver for agent actions. Claim a fresh desktop for the next task; Fleet doesn't revert snapshots.
4/ Ready to try it? The guide includes the image recipe and a Python sample for claiming the desktop, taking a screenshot, and calling Cua Driver. You'll need Fleet credentials, a registry image, and cua-sandbox from main. Docs: https://t.co/G7Lt9LKrXL Start a Fleet:
Context
Cua published a guide for running OSWorld tasks on Cua Cloud Fleets. OSWorld is a computer-use . The guide takes its official Ubuntu image and applications, adds Cua Driver, publishes the VM disk to a registry, and claims a desktop through the SDK.
This is a setup guide, not a new benchmark or a published score: Cua reports no result here. It uses OSWorld's setup tools and evaluators against the Fleet desktop, with Cua Driver handling the 's actions. Because Fleet does not revert snapshots, each new task needs a freshly claimed desktop. The guide needs Fleet credentials, a registry image and cua-sandbox from main.
- benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
- sandbox — An isolated environment where AI-generated code or agent actions run without being able to touch anything real.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Checking sign-in…
Loading comments…