Single-week surge of 2.5 times, 7-fold increase in 5 months! OpenAI’s Codex users have exceeded 8 million

```

OpenAI's AI programming tool Codex is experiencing a rare explosive growth in users. Since the launch of GPT-5.6, the combined weekly active users of Codex and ChatGPT Work surpassed 8 million within a few days, with weekly usage rising 2.5 times compared to before, and the total user base expanding more than 7 times in 5 months. The growth rate makes the entire enterprise SaaS industry pale in comparison.

OpenAI CEO Sam Altman confirmed on the social platform X on July 14 that the usage of the company's intelligent agent products (including Codex and ChatGPT Work) increased by 2.5 times in the past week.

Later, he posted again stating that the growth of GPT-5.6 Sol was "incredible," highly commending the reasoning team for their efforts to support the demand, and frankly admitting that some service fluctuations may occur recently.

The surge in users has directly impacted OpenAI's infrastructure, forcing competitor Anthropic to respond quickly—just hours after OpenAI announced crossing the 7 million user milestone, Anthropic extended Claude Fable 5's promotional pricing and increased Claude Code's weekly usage cap by 50%. The competitive landscape in the AI programming field is rapidly being reshaped.

GPT-5.6 Ignites the Growth Engine

Codex's user growth curve is almost a vertical line.

According to OpenAI Codex engineering lead Tibo Sottiaux, Codex had less than 1 million weekly active users in February this year, rising to 5 million by early June. After GPT-5.6 officially went live on July 9, the growth rate accelerated sharply: by July 12 it reached 6 million, broke 7 million about 24 hours later, and hit 8 million by Sunday.

Sottiaux announced this milestone on the social platform X and simultaneously announced resetting usage limits for all users, while maintaining the previous policy of lifting the 5-hour usage cap, encouraging users to explore the boundaries of GPT-5.6 Sol's capabilities.

The direct catalyst behind this growth was multiple product integrations completed by OpenAI on the same day: the standalone Codex app was merged into the ChatGPT desktop client, a new intelligent agent mode ChatGPT Work was launched for knowledge workers, and the Atlas browser began a phased shutdown. The centralized integration of products significantly lowered the user entry threshold.

Traffic Surge Reveals Scaling Limits

The influx of users immediately put OpenAI's infrastructure under pressure. According to company disclosures, traffic in the 48 hours after launch was about twice the previous peak, exposing several scaling issues.

On July 12, Sottiaux released a detailed note listing the team's responses: by optimizing reasoning, the available capacity per subscription user was increased about 10%; the context window was compressed from 372,000 tokens to 272,000 tokens (the previously larger limit had caused unexpected billing issues); the experimental reasoning intensity setting internally codenamed "juice" was rolled back; and overly aggressive multi-agent behavior at the highest reasoning levels was fixed.

Additionally, OpenAI temporarily lifted the 5-hour usage cap for Plus, Business, and Pro subscription users—this was the most open access policy since the product launched.

These adjustments sparked controversy in the community: some users regarded the reduction in context window as an indirect downgrade, while others positively evaluated Sottiaux's transparency in sharing operational trade-offs.

Sam Altman later weighed in, characterizing OpenAI's handling as a stance in opposition to competitors’ "disregard for users."

Platform Strategy: Tool Integration Becoming Core Competitiveness

Codex's explosive growth reflects a deeper product strategy shift at OpenAI. By integrating Codex, ChatGPT Work, and the built-in browser into a single desktop application, and connecting to mainstream tools like Slack, Google Drive, SharePoint, CRM, and calendars via plugin architecture, OpenAI is turning the ChatGPT desktop client into a unified workspace layered atop existing workflows.

This logic is also evident in the competition for cost efficiency. For example, Cognition’s Devin Fusion product, powered by Fable 5, can have lower task completion cost than Anthropic’s higher-priced Opus 4.8—because better task allocation reduces wasted work: in 81% of Fable-led runs, the main model never directly edited code. This case shows that if higher-priced models can effectively avoid resource waste, they might actually lower overall costs.

Usage Limits Remain a Growth Bottleneck

Despite strong growth momentum, usage limits are still the core bottleneck constraining user experience. Even though the 5-hour cap has been lifted, Codex and ChatGPT Work still share the same weekly usage quota pool. Once users run high-intensity Sol Ultra multi-agent orchestration tasks, the quota is consumed very quickly. Sottiaux has openly acknowledged this issue and said a fix is underway.

Sam Altman, while confirming growth data, conceded that as demand continues to surge, there may be fluctuations in service recently. Analysis points out that for developers and enterprise users investing in AI programming tools, in the second half of this growth race, infrastructure scalability may become as important as product capability.

Competitors Forced to Respond

It is noteworthy that the rapid rise in user numbers has directly impacted the competitive landscape.

Within hours of OpenAI announcing the breakthrough of 7 million users, Anthropic announced the extension of Claude Fable 5's promotional pricing to July 19, and a 50% increase in Claude Code’s weekly usage cap. The timing of these events overlaps closely, and the market's interpretation of their connection is obvious.

In performance testing, GPT-5.6 Sol has ranked second on the Arena agent leaderboard after 7,800 real-world intelligent agent task tests, with many developers once again seeing OpenAI as the leader in AI programming.

On the Anthropic side, the launch of its Claude Cowork product was likewise prompted by users’ spontaneous use of Claude Code in scenarios beyond software development, showing both companies are evolving in a similar direction regarding product form.

From a broader competitive perspective, OpenAI occupies a scale advantage with its massive base of 900 million weekly active users and leading benchmark scores; Anthropic retains deep accumulation among developers thanks to code quality reputation and business-friendly pricing; and for users who value data autonomy, open source solutions such as running GLM 5.2 or Kimi K2.7 offer more attractive choices at the expense of some speed.

Risk warning and disclaimerThe market has risks, invest with caution. This article does not constitute personal investment advice, nor does it take into account the individual user's specific investment objectives, financial situation, or needs. Users should consider whether any opinions, views or conclusions in this article are suited to their particular situation. Investing based on this is at your own risk. ```