Google Launches Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber

Google has introduced three new additions to the Gemini family of models: Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. These models are designed to provide the efficiency, latency, and reliability required for developers and customers to build and scale production AI agents. By optimizing the balance between quality and cost, these releases aim to improve the performance of agentic workflows across various technical and knowledge-based tasks.

Gemini 3.6 Flash: Enhanced Efficiency and Performance

Gemini 3.6 Flash serves as a workhorse model, building on feedback from the 3.5 Flash iteration. It delivers improvements in coding, knowledge work, and multimodal performance while significantly increasing token efficiency. According to the Artificial Analysis Index, the model reduces output token usage by 17% compared to 3.5 Flash. Furthermore, it requires fewer reasoning steps and tool calls to complete multi-step workflows.
The model is priced at $1.50 per 1 million input tokens and $7.50 per 1 million output tokens, offering a lower cost per agentic task. Performance gains are evident across several benchmarks, including DeepSWE, where it achieved 49% precision compared to 37% in 3.5 Flash, and MLE Bench, where it reached 63.9%. Additionally, it features built-in computer use capabilities via the Gemini API and Gemini Enterprise. To ensure safety, 3.6 Flash includes enhanced safeguards against chemical, biological, radiological, and nuclear (CBRN) risks and cyber offense misuses.

Gemini 3.5 Flash-Lite: Scaling High-Throughput Workflows

Gemini 3.5 Flash-Lite is the fastest model in the 3.5 series, engineered for low-latency tasks and high-throughput environments such as agentic search and document processing. It achieves 350 output tokens per second, as measured by the Artificial Analysis Index. With a price point of $0.30 per 1 million input tokens and $2.50 per 1 million output tokens, it provides a cost-effective solution for developers managing high-volume production traffic.
The model allows developers to configure thinking levels based on their specific workload needs, ranging from low-latency execution to more complex, multi-step subagent tasks. It demonstrates significant performance improvements over previous generations, outperforming 3.1 Flash-Lite across various agentic and coding evaluations. It also includes built-in computer use tools to support reliable task execution.

Specialized Security with 3.5 Flash Cyber

To address the growing need for efficient software security, Google has introduced Gemini 3.5 Flash Cyber. This model is fine-tuned specifically for detecting, validating, and patching cybersecurity vulnerabilities. It operates within the CodeMender code security agent, where multiple agents work in tandem to produce comprehensive security reports.
Due to the sensitive nature of this technology, 3.5 Flash Cyber will be available through a limited-access pilot program. It is intended for use by governments and trusted partners to help identify and fix critical vulnerabilities before they can be exploited. Beyond these releases, development continues on the Gemini 4 pre-training run, and Gemini 3.5 Pro is currently undergoing partner testing with plans for broader availability in the future.