According to CoinMeta, Google released its latest AI model on Wednesday, calling it a cutting-edge model with capabilities in coding, office work, and network defense. In the Deepswe v1.1 tests, this model scored 77.9%, outperforming Claude Opus with 74.2% and GPT-6 Astra with 74.1%. This model can also generate up to 1 million tokens in a single response, significantly improving processing power. Google stated that this model will be made available to vetted security teams through a program called Fairwind, which was launched on September 2nd and already has over 650 partners.