Google DeepMind releases Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber

Google DeepMind introduced three new models in the Gemini Flash series: Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. The lineup targets developers building production-grade AI agents by emphasizing token efficiency, lower latency, and reliable performance.

Gemini 3.6 Flash is positioned as a workhorse model, delivering improvements in coding, knowledge work, and multimodal tasks. According to the Artificial Analysis Index, it reduces output token usage by 17% compared to 3.5 Flash, and in some benchmarks like DeepSWE by Datacurve, reductions reach up to 65%. The model is priced at $1.50 per 1M input tokens and $7.50 per 1M output tokens, undercutting 3.5 Flash on cost per agentic task. Performance gains include higher precision in code edits (49% vs. 37% in DeepSWE), improved ML research capabilities (63.9% vs. 49.7% in MLE Bench), and stronger computer use performance (83.0% vs. 78.4% in OSWorld-Verified).

Gemini 3.5 Flash-Lite is described as the fastest and most cost-effective model in the 3.5 class, achieving 350 output tokens per second according to the Artificial Analysis Index. It is designed for high-throughput, cost-sensitive scenarios and significantly outperforms prior Flash-Lite generations in agentic workflows.

Gemini 3.5 Flash Cyber integrates a specialized cybersecurity-focused model with the CodeMender code security agent. The combination is positioned for competitive performance in cybersecurity applications that require careful orchestration of models and agent infrastructure.

The releases include Frontier Safety safeguards for Gemini 3.6 Flash, specifically targeting Chemical, Biological, Radiological, and Nuclear (CBRN) domains and cyber offense misuse resistance. These safeguards aim to reduce jailbreak susceptibility. Customers such as Hebbia and Harvey report Gemini 3.6 Flash as particularly capable for multimodal tasks like document parsing, chart and data analysis, and report drafting.