Skip to content

AI & Frontier Tech

Cognition Deploys GPT-6 Astra Across Devin Platform to Automate Code Verification and Simulator Testing

Autonomous software engineer runs live iOS game simulators and resolves visual customer bug reports with verifiable screenshots.

By Lazy Tensor
Published · 3 min read

Bay Area — Cognition deploys GPT-6 Astra to test Devin software engineer's code
Bay Area — Cognition deploys GPT-6 Astra to test Devin software engineer's code

Bay Area artificial intelligence developer Cognition is deploying OpenAI's newly released GPT-6 Astra model across its entire software engineering product lineup, including its flagship autonomous agent Devin, command-line interface tools, and desktop applications. Announced on Friday by engineering leaders in San Francisco, the integration focuses on automating end-to-end functional code verification, reducing manual human code review bottlenecks, and accelerating enterprise customer support workflows. Software development teams across San Francisco, San Ramon, Dublin, and Pleasanton have begun adopting the upgraded verification pipeline for production repositories.

In a technical demonstration illustrating the model's multimodal testing capabilities, Devin utilized GPT-6 Astra to autonomously test Otter Run, an interactive iPhone mobile game. Rather than relying solely on static unit tests, the agent launched the application inside an iOS simulator, played through interactive levels, and returned a full video recording of the game running alongside a structured verification report identifying checks that passed and edge cases left untested. Additionally, when an enterprise customer submits a screenshot of a user-interface bug, Cognition passes the image directly to Devin powered by Astra, which diagnoses the underlying code defect, implements the patch, and returns an updated screenshot proving the visual fix.

Walden Yan, Co-founder of Cognition, emphasized that automated proof of execution represents a major leap forward for autonomous software engineering teams. Yan stated that one of the big pieces that Astra improves on is its ability to test and prove that its work actually functions the way human engineers expect. Looking ahead to broader enterprise adoption, Yan added that his team expects over time that developers will have to manually look at less code and end up shipping more at the end of the day, calling it one of the capabilities Cognition is most excited about with GPT-6.

Enterprise engineering directors and technology board advisors in the Tri-Valley and Silicon Valley noted that shifting human developer attention from line-by-line syntax inspection to high-level architectural review and verifiable test artifacts substantially boosts software reliability. Quality assurance departments at regional financial and healthcare technology companies observed that automated video and screenshot evidence streamlines compliance auditing before production deployments.

Software engineers, CTOs, and developer productivity teams can view interactive demonstrations of Devin's GPT-6 Astra testing workflows, CLI documentation, and enterprise licensing tiers online at cognition.ai or by contacting Cognition's South Beach office on Second Street in San Francisco at (415) 610-8820. Technical webinars covering automated iOS simulator testing and continuous integration pipelines will run throughout November. Inquiries regarding enterprise sandbox trials may be directed to the solutions engineering department.

Comments (0)

No comments yet. Be the first to share a thought on this story.

The free newsletter

Silicon Valley, in your inbox.

The local stories worth knowing, delivered once a week.

From Silicon Valley, San Francisco, San Jose, Palo Alto, Mountain View, and Berkeley.

Get the newsletter

Free to read. Unsubscribe anytime.

More local stories