OpenAI is integrating its flagship GPT-6 Astra model and two smaller variants, GPT-6.1 Sol and GPT-6 Luna, into the Google Cloud Model Garden. This platform currently hosts over 200 neural networks from various providers. The models are expected to be generally available within the next few weeks.
The Sol and Luna models are optimized for cost-efficiency, priced 5 and 100 times lower than the flagship Astra model, respectively. Google Cloud customers can utilize existing long-term spending commitments to cover the costs of these models, simplifying the procurement process for enterprises.
Beyond the initial launch, the companies intend to integrate these models into Google’s Confidential Compute infrastructure, which utilizes Trusted Execution Environment technology to encrypt memory segments. This setup is intended to enhance security for sensitive workloads and collaborative data training.
OpenAI also plans to expand availability to Google’s Antigravity and Gemini Agent products. Antigravity is designed to distribute programming tasks among multiple agents, while Gemini Agent focuses on automating business workflows across various cloud applications.
