Google has launched new AI models designed to reduce latency and token costs for enterprise software agents, alongside a specialized security variant.

Google has introduced Gemini 3.6 Flash and Gemini 3.5 Flash-Lite, two new models engineered to lower latency and token costs for enterprise AI agents. The release aims to support high-volume, autonomous workflows by balancing reasoning capabilities with operational efficiency. Additionally, Google unveiled a restricted Gemini 3.5 Flash Cyber variant focused on vulnerability remediation, while integrating new computer-use tools directly into its API and enterprise platforms.

Performance and Pricing Metrics

Gemini 3.6 Flash features 17 percent fewer output tokens than the 3.5 Flash version, with pricing set at $1.50 per million input tokens and $7.50 per million output tokens. In benchmarks like DeepSWE, the model achieved a 49 percent success rate compared to 37 percent for its predecessor. Gemini 3.5 Flash-Lite is optimized for high-volume tasks, offering speeds of 350 output tokens per second at a cost of $0.30 per million input tokens and $2.50 per million output tokens.

Specialized Security and Integration

The Gemini 3.5 Flash Cyber variant is currently restricted to governments and vetted partners to prevent the generation of offensive exploit code. It is designed to validate and remediate code vulnerabilities, utilizing parallel instances to cross-check findings. Meanwhile, Google has integrated a client-side computer-use tool directly into its API, allowing models to operate on operating systems without custom intermediary software.