OpenAI Models Enhance Engineering Workflows Across Industries

OpenAI has announced advancements in how its models assist engineers in designing, simulating, and building physical products. These capabilities encompass mechanical and electrical engineering, as well as end-to-end system design, and are already aiding internal teams in testing ideas and accelerating iterations.

Engineers often need to integrate mechanical designs, electronics, software, and manufacturing requirements using specialized tools. OpenAI is training its models to facilitate this work, from preparing drawings and setting up simulations to coordinating changes across disciplines. The aim is to allow engineers more time to explore ideas and refine designs.

In the past year, enhanced capabilities of frontier models have enabled internal teams to test new concepts and iterate more quickly. Early indications suggest these advancements are also improving professional workflows in the industry. OpenAI is collaborating with partners to identify where these capabilities can have the most impact and what improvements are necessary. The company invites engineers, engineering firms, or software vendors involved in the design, simulation, manufacturing, or production of complex physical products to collaborate on shaping these tools and addressing specific challenges.

Designing a Device with Astra

OpenAI tested Astra’s performance in multimodal and computer use workflows by attempting to create a device without any direct human edits. Using GPT-6, the team defined functional requirements and designed the enclosure, printed circuit board (PCB), and software integrations necessary for a device that allows users to interact with Codex.

• Aesthetics and design: Astra developed the device’s appearance, drawing inspiration from a recent collaboration and references from games and toys.
• Mechanical design: Astra designed the enclosure and internal structures to secure the components.
• Electrical design: Astra selected electrical components and designed the PCB, including placements and connections to fit within the enclosure.
• UI/UX and software: Astra created the user interface and generated the firmware and software integrations needed to connect the device to Codex.

Two team members designed this test device with Astra in under a week as a side project, and a contract manufacturer produced a small batch of prototypes in less than a month.

Preparing the device for manufacturing required several rounds of review and iteration. Astra did not consistently account for the manufacturer’s production limits, tolerances, or assembly details, highlighting areas for future model improvements.

Accelerating OpenAI Hardware Teams

OpenAI’s hardware and infrastructure teams utilize Codex and ChatGPT to create and revise computer-aided design (CAD) models, prepare manufacturing drawings, set up simulations, and review designs across robotics and compute infrastructure. Since February, the usage of these models among hardware teams has increased 23-fold.

Measuring Progress on Engineering Tasks

OpenAI is also assessing progress in newer models on tasks that merge coding with engineering knowledge. BenchCAD 2.0 evaluates whether models can reconstruct 3D parts and assemblies and identify connections in circuit boards. A newer internal model achieved an average score of 0.875, compared to 0.594 for GPT-6 Astra at its highest reasoning setting. While the engineering lifecycle includes many elements beyond design, improvements in this area are seen as early evidence of AI models’ potential to enhance iteration speed in professional workflows.

Safely Deploying with Industry Collaborators

OpenAI is collaborating with a select group of industry partners to test these capabilities on real engineering challenges. Their engineers review model outputs for accuracy and practical use, providing insights into which tasks models can currently support and what improvements are needed for safe and reliable deployment.

The collaboration with Chip Ganassi Racing (CGR) aims to explore how OpenAI models can expedite engineering and prototyping processes. The R&D team is testing workflows that convert 3D scans of existing parts into editable CAD models and 2D manufacturing drawings, particularly useful when original design files are unavailable.

CGR engineers assess the models and drawings for accuracy and manufacturability, offering feedback that helps identify where models can minimize manual work and where further enhancements are necessary.

Additionally, OpenAI is announcing a new research collaboration with Mazda, where its research and Forward Deployed Engineering teams are working with Mazda’s aerodynamics and crash-safety engineers to explore how OpenAI models can aid in developing and evaluating new vehicle designs.

This collaboration uses simulations of airflow and structural behavior to investigate performance improvements while considering styling and cost. Further progress and results will be shared in the coming months.

GPT-6 Astra assists Mazda’s aerodynamics engineers in exploring ways to reduce vehicle drag by modeling new nose shapes, simulating airflow, and refining the most promising designs. Agreement for use has been secured from Mazda Motor Corporation, with any republication or reuse elsewhere strictly prohibited.

Help Shape AI for Engineering and Manufacturing

OpenAI has been collaborating closely with industry partners and early adopters to expand the capabilities of AI agents in hardware and manufacturing. These initial results show promising potential, and OpenAI is seeking additional collaborators to accelerate capability improvements and ensure safe deployment in the industry.

Those interested in shaping the training and development of frontier models by identifying high-value use cases, software tooling, or future opportunities in mechanical engineering, electrical engineering, electromechanical design, mechatronics, or related fields are encouraged to fill out a collaboration form.

 

Source: OpenAI News