Google Grants Engineers Access to Claude Opus 5 Through Antigravity Platform

How Claude Opus 5 Fits Into Google's Internal AI Evaluation Process

Google has provided all of its engineers with access to Anthropic's Claude Opus 5 model via its internal Antigravity system, according to a report from Business Insider. The move, confirmed in September 2026, allows engineers across teams to experiment with the advanced AI model while the company maintains that its own Gemini series remains the primary and foundational model for its products and research. The access is being rolled out as part of Google's broader strategy to evaluate competing AI systems internally.

The Antigravity platform, Google's internal tool for testing and integrating external AI models, now includes Claude Opus 5 as an available resource for engineering staff. This enables teams to benchmark performance, explore use cases, and assess capabilities outside of Google's proprietary ecosystem. Despite the access, Google emphasized that Gemini continues to serve as the core model powering its consumer-facing applications, cloud services, and long-term AI roadmap. The company frames the initiative as a way to stay informed about advancements in the field without shifting its strategic focus.

What Does This Mean for Google's AI Strategy Moving Forward?

Engineers are using Claude Opus 5 through Antigravity to test The model is not being integrated into public products but is instead used for comparative analysis against Gemini and other internal prototypes. Google says this approach helps identify strengths and weaknesses in competing models while protecting its own technological direction. Access is granted uniformly across engineering roles, with usage monitored for security and compliance purposes.

By allowing engineers to work with external models like Claude Opus 5, Google aims to maintain a broad understanding of the AI landscape while reinforcing confidence in its own Gemini architecture. The company states that exposure to diverse systems improves internal innovation and helps anticipate industry trends. However, it reiterates that no external model will replace Gemini as the foundation of its AI efforts. This balanced approach reflects Google's effort to remain both open-minded and strategically disciplined in a rapidly evolving field.

Is Google planning to replace Gemini with Claude Opus 5? No, Google has explicitly stated that Gemini remains its primary and foundational AI model, and access to Claude Opus 5 is strictly for internal evaluation and benchmarking purposes.

Frequently Asked Questions

Can engineers use Claude Opus 5 in public-facing projects? No, the model is only accessible through the Antigravity platform for internal testing and is not approved for deployment in consumer products or services.

How does Google ensure security when using external AI models internally? Access to models like Claude Opus 5 via Antigravity is monitored, restricted to approved use cases, and subject to the company's internal AI safety and data protection policies.