NEWS

On 20 real engineering jobs, the best AI model scored 37%; five models scored zero.

Xplore AI's new RWE-bench tested 11 AI models on 20 engineering tasks, including crankshaft design, DDR3 board routing and turbine vane CFD, each graded by a deterministic verifier with no human or AI judge. The top model, GPT-6 Astra, averaged a 37% pass rate. The hard design work is still mostly done by people.

Xplore Real World Engineering Bench chart of frontier AI model scores on 20 tasks
Image: Xplore AI

More from The Feed

NEWS

At Holcim, SAP maintenance plans now dispatch robot dogs and drones, and findings come back as SAP notifications.

SAP unveiled Physical AI at TechEd, and at Holcim a maintenance plan in SAP Plant Maintenance now triggers inspections by quadruped robots and industrial drones of conveyor belts and kilns. Once the inspection runs, a maintenance notification is created back in SAP. Inspection is moving inside the maintenance workflow instead of beside it.