Google Relents on Internal Model Access, Opens Door to Rival Anthropic's Claude

Google has made a significant internal policy shift by opening access to Anthropic's Claude Opus 5 model to all engineers across the company. This move marks a departure from its long-standing requirement for employees to prioritize its own Gemini model for internal development.

The decision reflects the intensifying competition in AI coding tools and acknowledges that Gemini still lags behind frontier models like Claude in programming capabilities. Google's Gemini model, despite recent updates, has been criticized by some engineers for its limitations on complex coding tasks.

According to a Google spokesperson, engineers can access select third-party models, including Anthropic's Opus 5, through the company's internal development platform, Antigravity. However, employees cannot directly use Claude Code and must instead select the Opus 5 model within the Antigravity platform.