OpenAI Launches Codex on Mobile to Compete with Anthropic

The Rise of AI in the Professional World

In 2026, artificial intelligence continues to transform work methods, particularly with AI agents capable of performing tasks on behalf of users. OpenAI, known for ChatGPT, recently announced a significant advancement for its Codex tool, now available on mobile in beta starting today. This application allows users to control the AI agent remotely, even while on the go.

One of the major benefits of AI is its ability to execute concrete tasks on a computer, thereby freeing up the user. The success of Claude, the chatbot from Anthropic, is largely attributed to its agents like Claude Caude and Claude Cowork, which have won over businesses and professionals. Meanwhile, OpenAI has developed Codex, a similar tool primarily aimed at developers, but which also proves effective for automating administrative tasks.

Codex: A Versatile Tool

At OpenAI, Codex is used by nearly all departments to automate work. For example, it has processed 24,771 tax forms, totaling 71,637 pages. One employee even reported saving between 5 and 10 hours per week by generating weekly reports using Codex. This versatility makes Codex indispensable for many tasks within the company.

The mobile version of Codex does not replace the AI agent on the computer, but it allows for remote control. Users can launch new projects, review the agent's results, guide its work, and validate its conclusions, all without being physically present in front of their computer. With this application, Codex users can continue to leverage AI even when they are on the subway, during lunch breaks, or at the gym.

OpenAI Facing Competition from Anthropic

Although ChatGPT is extremely popular with nearly a billion active users per month, OpenAI faces increasing competition from Anthropic, whose AI Claude is increasingly appealing to businesses. According to the Ramp AI index, Anthropic has recently surpassed OpenAI in terms of adoption by American companies, with 34.4% compared to 32.3% for OpenAI. These results are based on measuring AI adoption within 50,000 American companies.

This trend is explained by the rapid increase in the adoption of Anthropic's models, which has quadrupled in a year, while OpenAI's progress has only been 0.3%. Furthermore, Anthropic could soon surpass OpenAI in terms of valuation, thanks to a potential fundraising round of $30 billion, bringing its valuation to $900 billion according to the Financial Times.

OpenAI has clearly directed its efforts towards the enterprise segment, a particularly lucrative market. The development of new products and features, such as the GPT-5.5 model, reflects this focus. The latter allows OpenAI to catch up with Anthropic on the most important productivity criteria. Additionally, GPT-5.5 Instant, the new default AI offered for free, has addressed certain flaws, such as unnecessary phrases, while reducing hallucinations.