October 1, 2026
Google is facing internal doubts about the efficiency and ability of its new artificial intelligence model, “Gemini 4,” to compete in key areas such as programming, as the company prepares to launch it.
Although the model successfully passed the criteria for evaluating the efficiency of artificial intelligence, informed sources indicated to Bloomberg that its actual performance and practical use by employees show less efficiency, especially in dealing with programming tasks, which raised internal questions about its competitive ability in front of the companies “Open AI” and “Anthropic”. It is noteworthy that Google was planning to launch a previous version called “Gemini 3.5 Pro” last June before abandoning that step.
On the other hand, Google rejects the allegations of Gemini 4’s poor performance in programming, referring to the statements of the head of Google DeepMind, Koray Cavukoglu, who expressed during a conference for “The Information” website his absolute confidence in the team and his assurance that the company will always remain at the forefront. One of the knowledgeable employees also denied the presence of any difficulties facing the models in complex programming tasks in practice, stressing that there is a broad consensus that the model represents the latest technology after conducting careful tests on it.
Opinions vary within the company; One team of employees believes that competing models such as “Fable” from “Anthropic” and “Astra” from “OpenAI” are developing at a faster pace, while another team believes that the upcoming version has succeeded in catching up with leading artificial intelligence laboratories.
The successes of “Gemini 4” are of great strategic importance to Google, given the reliance of a wide sector of its core products on it, such as artificial intelligence answers in the search engine (the main source of the company’s profits), in addition to Google Maps, Gmail, and the Chrome browser, which are services that each serve more than a billion users, giving Google a wide spread advantage compared to its competitors.
With OpenAI and Anthropic moving toward developing their own software and products rather than just selling models, any failure by Google to provide a sophisticated model may give its competitors a greater opportunity to convince consumers, developers, and companies to adopt their platforms in the future of search and software.
