Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber

Google Releases Trio of Specialized Gemini Models for AI Agents

Google has announced an expansion of its Gemini family with three new models designed to address the growing demand for efficient, scalable AI agents. The release includes Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and the specialized Gemini 3.5 Flash Cyber. This move signals a clear focus on providing developers and enterprises with a range of tools optimized for the performance, latency, and cost requirements of production-grade agentic workflows, moving beyond a one-size-fits-all approach to AI development.

New Flash Models Target Efficiency and Scale

The new models are engineered to provide a balance of capability and operational efficiency. Google is emphasizing significant improvements in token usage, speed, and cost-to-performance ratios, directly addressing feedback from developers building complex, multi-step AI systems. The company also announced that Gemini 3.5 Pro is in testing with partners and that pre-training has begun for the next-generation Gemini 4.

• Gemini 3.6 Flash: Positioned as the new workhorse model, it reportedly reduces output token usage by 17% compared to 3.5 Flash. It is priced at $1.50 per 1M input tokens and $7.50 per 1M output tokens, offering improved performance in coding (49% on DeepSWE) and multimodal tasks at a lower overall cost.
• Gemini 3.5 Flash-Lite: This is the fastest and most cost-effective model in the 3.5-series, delivering 350 output tokens per second. Priced at $0.30 per 1M input tokens and $2.50 per 1M output tokens, it is built for high-throughput tasks like agentic search and document processing.
• Gemini 3.5 Flash Cyber: A specialized model fine-tuned for finding and fixing cybersecurity vulnerabilities. It will be available exclusively to governments and trusted partners via the CodeMender agent in a limited-access pilot program.

Market Implications: The Race for Agentic Dominance

This launch intensifies the competition to provide the foundational models for the emerging agent economy. By offering a tiered portfolio of models—from the high-speed, low-cost Flash-Lite to the more capable 3.6 Flash—Google is directly catering to the practical need for economically viable AI applications. The strategy puts pressure on competitors to offer similarly granular and cost-effective solutions for developers who are moving from experimentation to production. The restricted, security-focused release of Gemini 3.5 Flash Cyber also highlights the industry's cautious approach to deploying powerful, dual-use AI capabilities, establishing a controlled access model for sensitive applications.

Google's multi-model release underscores a critical industry trend: the future of AI is not a single, monolithic model, but a diversified portfolio of specialized, cost-efficient models purpose-built to make complex agentic systems economically viable at scale.