OpenAI launched GPT-5.6 last week and folded Codex into a unified ChatGPT desktop app, igniting a period of intense activity for the artificial intelligence leader. In the wake of these significant product integrations and updates, the company has witnessed an unprecedented surge in user adoption, fundamentally reshaping the competitive dynamics within the AI development and knowledge work sectors. Early Tuesday, Tibo Sottiaux, engineering lead for Codex at OpenAI, shared insights on X (formerly Twitter) indicating that the combined active users of Codex and ChatGPT Work might be approaching the 8 million mark. This trajectory represents a remarkable acceleration, particularly in the context of the significant growth observed by competitors such as Anthropic’s Claude throughout 2026.
The user growth figures for OpenAI’s integrated offerings paint a striking picture of rapid adoption. Codex, which boasted fewer than 1 million weekly active users in February, had already climbed to 5 million by early June. The launch of GPT-5.6 on July 9th served as a catalyst, dramatically accelerating this upward trend. Within just three days, by July 12th, the user count had reached 6 million. The following day saw it climb to approximately 7 million, and by Sunday, the combined active user base had reportedly hit the 8 million milestone. This growth curve is virtually unheard of for enterprise SaaS products, which typically experience much slower adoption rates over years, if at all.
A Tumultuous Launch and Strategic Integration
The recent surge in user activity follows a series of significant strategic moves by OpenAI, executed concurrently. The company integrated its standalone Codex application into the unified ChatGPT desktop app, simultaneously introducing ChatGPT Work as a new agentic mode specifically designed for knowledge workers. Concurrently, OpenAI began the process of sunsetting its Atlas browser. This multi-pronged launch, occurring within a single day, placed immense pressure on OpenAI’s infrastructure and operational capabilities.
According to company statements, demand surged almost immediately following the launch, with traffic roughly doubling OpenAI’s previous peak within the first 48 hours. This unprecedented influx exposed several scaling challenges. In a detailed thread published on July 12th, Sottiaux outlined the engineering team’s swift response to these issues. The measures implemented included optimizing inference processes to enhance capacity by approximately 10% per subscriber. To address unintended billing complexities arising from the extended context window, the team reduced the token limit from 372,000 to 272,000. Furthermore, experimental reasoning-effort settings, internally referred to as "juice" values, were rolled back, and overly aggressive multi-agent behaviors at the highest reasoning levels were patched.
In a move that offered users unprecedented access, OpenAI also temporarily removed the five-hour usage cap for its Plus, Business, and Pro subscribers. This decision represented the most generous access period offered by the product since its inception, signaling a strategic effort to accommodate the overwhelming demand and retain newly acquired users.
Community reactions to these operational adjustments were varied. Some users interpreted the context window rollback as a subtle product downgrade. Others, however, commended Sottiaux for his transparent communication regarding the operational trade-offs involved. OpenAI CEO Sam Altman weighed in on the situation, posting on X a statement that positioned OpenAI favorably against competitors, framing their approach as one that respects user needs, in contrast to those he characterized as treating users with "contempt."
Competitive Landscape Responds to OpenAI’s Momentum
The rapid user growth experienced by OpenAI did not go unnoticed by its key competitors. Within hours of OpenAI announcing its 7 million user milestone, Anthropic took decisive action. The company announced an extension of its promotional pricing for Claude Fable 5 through July 19th and simultaneously increased the weekly usage limits for Claude Code by 50%. While the direct causal link between these competitive moves and OpenAI’s surge remains speculative, the timing and nature of Anthropic’s adjustments suggest a strategic response to maintain market share and user engagement.
The impact of GPT-5.6 on the developer community is also becoming evident. Since Thursday, GPT-5.6 Sol has ascended to the second position on Arena’s agent leaderboard, accumulating over 7,800 real-world agentic sessions. This climb has led many developers to re-evaluate OpenAI’s position as the leader in AI-powered coding assistance.
The Crucial Metric: Cost Per Task
In the competitive arena of AI development tools, the cost per completed task is emerging as a critical differentiator. Developers are increasingly judging AI systems not solely on their raw capabilities but on their economic efficiency in achieving desired outcomes. Cognition, for instance, recently reported that its Devin Fusion product, powered by Anthropic’s Fable 5, can prove more cost-effective per completed task than Anthropic’s more premium Opus 4.8. This cost advantage is attributed to Fable 5’s superior delegation capabilities, which reportedly reduce unnecessary computational work. According to Cognition’s data, in 81% of runs led by Fable, the primary model did not need to edit code, indicating a more efficient and accurate initial output. This highlights a key principle: a more expensive model can still lead to lower overall costs if it significantly minimizes wasted effort and rework.
The "Harness" as the Product: A Paradigm Shift
OpenAI’s strategic decision to consolidate Codex, ChatGPT Work, and its integrated browser into a single desktop application signifies a broader trend in the AI industry. The company is increasingly focusing on creating a comprehensive "harness" that integrates seamlessly with existing user workflows. This approach is further exemplified by OpenAI’s robust plugin architecture, which facilitates connections to a wide array of popular business tools, including Slack, Google Drive, SharePoint, Customer Relationship Management (CRM) systems, and calendar applications. By doing so, OpenAI is effectively transforming its AI offerings into a central workspace that complements and enhances the tools users already rely on.
Anthropic appears to be following a similar strategic path with its Claude Cowork offering. This initiative reportedly stemmed from the company observing that users were increasingly employing Claude Code for tasks extending far beyond traditional software development, indicating a natural evolution toward broader knowledge work applications.
Usage Caps: The Lingering Bottleneck
Despite the impressive growth and strategic advancements, usage caps remain a significant bottleneck for many users. Even with the temporary lifting of the five-hour limit for certain subscriber tiers, Codex and ChatGPT Work are still constrained by a shared weekly usage pool. For users engaging in intensive tasks, such as running heavy Sol Ultra sessions with complex multi-agent orchestration, this allowance can be consumed extremely rapidly. Sottiaux has acknowledged this challenge and indicated that a solution is currently under development, signaling a commitment to addressing this critical user pain point.
In the broader competitive landscape, OpenAI commands significant gravity with its massive user base and strong benchmark performance. Anthropic, however, continues to benefit from substantial developer goodwill, bolstered by its reputation for producing high-quality code and offering enterprise-friendly pricing structures. For developers who prioritize data control and are willing to trade some speed, open-source stacks running models like GLM 5.2 or Kimi K2.7 present increasingly attractive alternatives. The proliferation of coding agents and AI-powered knowledge work tools is accelerating at an astonishing pace, suggesting that the current competitive dynamics are likely to continue evolving rapidly. The ongoing innovation in this sector underscores the transformative potential of AI in reshaping how individuals and organizations work and develop.
