Even Google with Gemini Turns to Claude for Coding... "Competition Intensifies"

Google Grants Employees Access to Claude
"Gemini Remains Primary; External Models Available with Allocation Limits"

Google has allowed its employees to use "Claude," an artificial intelligence (AI) model developed by rival company Anthropic. Although this decision aims to increase work efficiency, concerns have been raised that Google’s own in-house model, Gemini, may be sidelined as a result.

On September 14 (local time), U.S. media outlet Business Insider reported that Google recently enabled its engineers to use Anthropic's coding model through its internal development tools. Google employees perform coding work on the company's internal development platform called "Antigravity," and going forward, Antigravity will support Anthropic’s Claude Opus 5 model.

Artificial Intelligence (AI) Model. Photo by AFP News Agency

Up to now, Google has not used coding models from competitors such as Anthropic or OpenAI. Instead, the company mandated the use of Gemini, with other external models being available only to specific teams such as Google DeepMind, the AI research subsidiary.

Business Insider noted, "This move demonstrates how fierce the AI coding competition has become," attributing this to Google's decision to allow the use of a rival company’s model, even after having invested astronomical sums in developing Gemini.

However, Google stated that "engineers have already been able to use a limited number of external models within Antigravity," and clarified that "Gemini remains the leading and foundational in-house model, while models from external companies are only offered within defined quotas for specialized use cases." In other words, core work continues to be performed using Gemini, while other models are deployed only to supplement specific tasks.

Nevertheless, some have suggested that Gemini’s capabilities may be lacking. According to the report, Google recently released the Gemini Flash 3.8 model with improvements across several functions, including coding tasks, but it is still considered to lag behind Anthropic and OpenAI.

Meanwhile, not only Google, but other big tech firms are also allowing the use of AI models developed by competing companies. For example, Amazon, which has its own "Nova" model, reportedly permitted employees to use models such as Claude from other firms earlier this year as well.

This content was produced with the assistance of AI translation services.