OpenAI: ChatGPT and Codex accelerate engineering for the physical world
On 3 Nov 2026 OpenAI shared progress on using its models to help engineers design, simulate, and build physical products across mechanical and electrical work; internal hardware-team model usage has grown about 23× since February, with industry pilots including Chip Ganassi Racing and Mazda.
Verified against OpenAI’s post Accelerating engineering for the physical world (dated November 3, 2026): OpenAI describes how its models help engineers design, simulate, and build physical products, spanning mechanical and electrical engineering and end-to-end system design, and already helping internal teams test ideas and iterate faster.
OpenAI says engineers must combine mechanical designs, electronics, software, and manufacturing requirements in specialized tools. It is teaching models to support that work—from preparing drawings and setting up simulations to coordinating cross-discipline changes—so engineers can spend more time exploring ideas and improving designs. Over the past year, stronger frontier models have helped internal teams move faster, with early signs of benefit in professional industry workflows; OpenAI is working with collaborators to learn where the gains are largest and what still needs to improve.
Designing a device with Astra (internal test)
The team used GPT-6 to define functional requirements and design the enclosure, printed circuit board (PCB), and software integrations for a device people can use to talk to Codex, testing Astra on multimodal and computer-use workflows—including whether a device could be created without direct human CAD edits. Astra contributed aesthetics, mechanical enclosure and internals, electrical component selection and PCB layout, plus UI/firmware. Two team members finished the side project in under a week; a contract manufacturer built a small prototype batch in under a month. OpenAI also states that manufacturing readiness still needed several human review rounds, and Astra did not consistently account for production limits, tolerances, or assembly details—gaps that will guide future model work.
Internal hardware teams
OpenAI’s hardware and infrastructure teams use Codex and ChatGPT to create and revise CAD, prepare manufacturing drawings, set up simulations, and review designs across robotics and compute infrastructure. Since February, model usage on those hardware teams, measured in tokens, has grown about 23-fold.
BenchCAD 2.0
BenchCAD 2.0 tests whether models can reconstruct 3D parts and assemblies and identify circuit-board connections. OpenAI reports a newer internal model scored 0.875 on average versus 0.594 for GPT-6 Astra at its highest reasoning setting. It notes the engineering lifecycle includes much more than design, and treats these gains as early evidence that models may speed professional iteration.
Industry collaborators
- Expanded work with Chip Ganassi Racing (CGR) on workflows that turn 3D scans of existing parts into editable CAD and 2D manufacturing drawings when original files are missing; CGR engineers review accuracy and manufacturability.
- A new research collaboration with Mazda, working with aerodynamics and crash-safety engineers on airflow and structural simulations that also consider styling and cost; more results are expected in the coming months.
OpenAI invites engineers, firms, and software vendors in mechanical, electrical, electromechanical, and related fields to apply to collaborate. This is a capability-progress and partnership update; it does not announce a new generally available model version or a pricing-table change.