Google DeepMind has launched three new Gemini AI models: Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. The 3.6 Flash serves as the flagship workhorse, enhancing coding, knowledge work, and multimodal tasks while reducing token consumption by up to 17%, making it more cost-effective than the prior 3.5 Flash. The Flash-Lite version focuses on affordability, targeting users who need a budget-friendly option. Meanwhile, Flash Cyber is a cybersecurity-specialized model, fine-tuned to identify and remediate vulnerabilities, but it will be available solely to government entities and select trusted partners through a limited pilot program.

These releases emphasize Google's commitment to improving efficiency, latency, and reliability for large-scale AI agent applications. The company aims to support developers building AI solutions by optimizing performance and cost, particularly for production environments requiring speed and precision. Notably, the 3.5 Flash Cyber model represents Google's strategic move into cybersecurity-specific AI tools, reflecting growing industry demand for specialized applications of artificial intelligence.

However, the announcement also underscored what was missing from the update: the much-anticipated Gemini 3.5 Pro. This premium model, known for its superior capabilities in complex reasoning and coding, remains unreleased since February’s last update. Competing AI companies, including OpenAI and Anthropic, have advanced rapidly with GPT-5.5, GPT-5.6, and new Claude models, intensifying pressure on Google to expedite its flagship offering. Industry reports suggest internal challenges at Google have delayed the Pro model’s launch due to unmet performance goals.

Google DeepMind’s product lead, Logan Kilpatrick, indicated that testing for Gemini 3.5 Pro is ongoing with partners, and an official rollout is expected soon. Additionally, the organization has initiated its most ambitious pre-training process for the upcoming Gemini 4, signaling continued investment in next-generation AI capabilities. These developments demonstrate Google's active strategy to remain competitive in the fast-evolving AI landscape despite temporary setbacks in delivering its top-tier model.