Back to feed
News Story
OpenAI Developers (X)
2 sources

7M+ weekly Codex users. 150+ updates in two months. @romainhuet catches you up o…

Original

OpenAI's Codex now has over 7 million weekly users and has rolled out 150+ updates in two months, including GPT-5.6, /goal, AppShots, Inline edits, Sites, mobile and SSH workflows, and improved PR workflows.

SynthePulse Insight · AI deep reading

Codex's Evolution: From Code Assistant to Development Partner

Version 1 · 2 sources

OpenAI has rolled out over 150 updates in two months, Codex's weekly active users surpass 7 million, and new capabilities include GPT-5.6, Ultra mode, multi-agent parallel collaboration, transforming it from a "code-writing tool" into an end-to-end development partner.

  • Codex's weekly active users exceed 7 million, with over 150 updates in two months.
  • New GPT-5.6 and Ultra mode support more complex reasoning and task handling.
  • The /goal command enables multi-agent parallel collaboration, handling multiple subtasks simultaneously.
  • Computer and browser operation speeds improved, with support for app screenshot understanding (AppShots) and inline editing.
  • New features include Sites one-click deployment, cross-project management, Codex mobile, and SSH workflows.
  • PR workflow covers the full process from code review to merge.
Open section navigationUser Scale and Update Pace

User Scale and Update Pace

According to a post by the OpenAI Developers official account on July 14, 2026, Codex's weekly active users have exceeded 7 million, and over 150 updates have been rolled out in two months. This data was summarized by OpenAI's Developer Relations lead Romain Huet in a video.

Such a high frequency of iteration indicates that Codex is in a rapid growth phase, with user base and feature density rising in tandem.

Core Model Upgrades: GPT-5.6 and Ultra Mode

The two most significant updates are GPT-5.6 and Ultra mode. GPT-5.6 is a new version of Codex's underlying model, while Ultra mode likely represents a higher-performance, more compute-intensive operating state for handling complex tasks.

Although official specific performance metrics have not been released, combined with the description of "faster computer use," it can be inferred that Ultra mode achieves breakthroughs in reasoning speed and task complexity.

Multi-Agent Parallelism and Automated Workflows

The new /goal command supports multi-agent parallel collaboration, allowing Codex to handle multiple subtasks simultaneously, thereby breaking down complex projects. This feature upgrades Codex from a single-threaded code generation tool to a development partner capable of coordinating multiple tasks.

The PR workflow covers the full process from code review to merge, and combined with Sites one-click deployment, Codex can now participate in the end-to-end pipeline from coding to deployment. Additionally, AppShots enables Codex to understand app screenshots, and Inline edits support in-line code and documentation modifications.

Cross-Platform and Mobile Expansion

The addition of Codex mobile and SSH workflows allows developers to code on mobile devices and connect to remote servers via SSH. This expands Codex's usage scenarios beyond the desktop IDE.

Cross-project management capabilities allow Codex to coordinate across multiple projects, further strengthening its role as a "development partner."

Positioning Shift: From Tool to Partner

Taken together, all updates show that Codex is no longer just a tool that "helps you write code," but a development partner capable of breaking down complex tasks, operating and testing applications, coordinating multiple projects, handling bugs and code reviews, and assisting with releases.

This shift means that AI programming tools are evolving from auxiliary coding to full-process automation, but the specific effectiveness and reliability still need to be verified by more real-world cases.

Credibility boundary

This article's information primarily comes from OpenAI's official developer account X posts and Chinese community translations. Core data such as user scale, update count, and feature list come from official statements but lack independent third-party verification. Some feature descriptions (e.g., specific capabilities of Ultra mode) lack detailed technical explanations and are inferential interpretations.

Insight takeaway

Codex has evolved from a code assistant to a development partner covering the full pipeline of coding, review, and deployment through over 150 updates in two months, with over 7 million weekly active users. However, the actual effectiveness and reliability of new features remain to be observed.