OpenAI says GPT-6 Astra designed a hardware device from scratch in under a week, but required revisions for manufacturing
OpenAI reports its model handled everything from enclosure to firmware for a new device, showing rapid AI-driven hardware iteration while highlighting gaps in manufacturing adaptability.
According to The Decoder, OpenAI stated that its hardware team used GPT-6 Astra to design a device allowing users to interact with the Codex coding agent, completing the task from scratch in under a week. The model handled every step, including enclosure design, PCB layout with component selection and trace routing, firmware, and the user interface.
Despite the speed of the design phase, a contract manufacturer took several weeks to produce a small batch of prototypes, requiring multiple revision rounds because Astra did not consistently account for manufacturing constraints, tolerances, and assembly details. This case illustrates that while frontier large models can significantly compress conceptual design and initial engineering time, human oversight regarding process feasibility remains essential when translating designs into manufacturable physical products.
Additionally, OpenAI noted that a newer "internal model" scored 0.875 on the BenchCAD 2.0 benchmark, significantly higher than GPT-6 Astra's 0.594 at its highest reasoning level. This benchmark tests the ability to reconstruct 3D parts and identify connections on PCBs. OpenAI has not shared details about this internal model.