Alphabet has introduced three new Gemini models, expanding its AI portfolio with a stronger focus on enterprise customers, cybersecurity and lower operating costs.

The launches include Gemini 3.5 Flash Cyber, a security-focused model designed to identify and patch software vulnerabilities, alongside updated Flash and Flash-Lite models aimed at improving efficiency for developers and businesses.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

Cybersecurity Becomes A Bigger Focus

Gemini 3.5 Flash Cyber will initially be available through a pilot program for governments and selected partners. Google said the model is priced below larger alternatives on a per-token basis, reducing costs for organisations running security workloads at scale.

The launch comes as AI companies increasingly compete in automated software security. Anthropic has already entered the segment with its Mythos model, making cybersecurity another key area of competition among foundation model providers.

Lower Costs, Higher Efficiency

Google also unveiled Gemini 3.6 Flash, which improves coding, multimodal and knowledge-work capabilities while using up to 17% fewer tokens than the previous version.

The company is also expanding Gemini 3.5 Flash-Lite, its fastest and lowest-cost model in the 3.5 family. It is intended for high-volume workloads and smaller tasks within AI agent systems, where cost and speed are often as important as model performance.

Competition Continues To Intensify

The announcements come ahead of Alphabet’s earnings report and as Chinese AI developers continue to expand their presence in the market.

Moonshot AI recently limited new subscriptions and API access for its Kimi K3 model after demand exceeded available capacity. Alibaba is also promoting its Qwen 3.8 Max model as one of the strongest performers in the sector.

The rapid growth in demand has increased pressure on AI companies to expand computing capacity alongside model development.

Alphabet Leans On Its Infrastructure

Google’s vertically integrated approach gives it greater control over AI infrastructure through its custom chips, cloud platform and in-house model development.

The company is also reportedly developing a new chip that could run Gemini models up to 10 times more efficiently. A Google Cloud spokesperson said the company continues to explore new hardware and software designs to improve performance and reduce the cost of serving AI models.

Gemini Road Map Advances

Google also provided an update on its broader Gemini roadmap. Gemini 3.5 Pro is now being tested with selected partners ahead of a wider release, while the company has begun what it describes as its largest pretraining run yet for Gemini 4.