OctoLink GEO

OpenAI Codex: Desktop App Launch, Enterprise Expansion, Hardware Collaboration & Technical Fixes

Author Editor
OpenAI Codex: Desktop App Launch, Enterprise Expansion, Hardware Collaboration & Technical Fixes

OpenAI integrates Codex with ChatGPT in a new desktop app, expands enterprise use cases with Samsung and Australian Payments Plus, launches a Codex...

OpenAI Codex ChatGPT AI Enterprise AI AI Programming Hardware Nvidia Meta Samsung Australian Payments Plus

OpenAI launched a new desktop application on July 10 that integrates Codex and ChatGPT experiences, enhancing AI programming capabilities while extending access to non-developer users. This move reflects the company’s push to make AI tools more inclusive beyond the developer community.

Enterprise adoption of Codex has reached new heights: Samsung Electronics rolled out ChatGPT and Codex to all its employees globally and those in its Device eXperience (DX) division, marking one of OpenAI’s largest enterprise deployments to date. Australian Payments Plus also leveraged ChatGPT Enterprise and Codex to boost efficiency across various business processes, sharing insights into scaling AI applications.

In hardware news, OpenAI announced a collaboration with Work Louder to release a Codex-branded keyboard on July 15. This dedicated input device is tailored to support AI development workflows for Codex users.

Technical issues were addressed recently: OpenAI fixed an anomaly where Codex user quotas were consumed abnormally quickly, caused by backend computational overload and abnormal auto-audit operations. The company reset all user quotas and added new monitoring mechanisms. Additionally, the quota reset cycle for free and Go users was adjusted from weekly to monthly, while Team accounts retain the 7-day cycle. A promotion allowed users to reset their usage by inviting friends over a two-week period.

Other key updates include OpenAI’s acquisition of Ona, a startup whose technology strengthens Codex’s enterprise capabilities—particularly in managing infrastructure, data, and security boundaries. Nvidia’s Polar framework boosted Codex’s performance by 594.74% using generalized relative policy optimization (GRPO) training without disrupting existing tool usage. Meta restricted its AI engineers from using Codex and Claude to avoid unintended model distillation, which violates user agreements.

Sources

  • iT Home Codex News Collection (https://www.ithome.com/tags/Codex/)

Related reading