Google Releases New Gemini Models (3.6 Flash, 3.5 Flash-Lite, 3.5 Flash Cyber)

Google has introduced new models in the Flash series, which aims to find the optimal balance between efficiency and quality for scaling AI agent workflows.

The key models released are as follows:

• 3.6 Flash: The flagship model with improved coding, knowledge work, and multimodal performance, reducing output token usage by 17% compared to 3.5 Flash, with reductions of up to 65% on some benchmarks.
• 3.5 Flash-Lite: The fastest and most cost-efficient model, generating 350 output tokens per second.
• 3.5 Flash Cyber: Combined with the security agent CodeMender to deliver specialized performance for cybersecurity.

In particular, 3.6 Flash delivers higher precision at a lower cost ($1.50/1M input, $7.50/1M output) than 3.5 Flash. It has demonstrated performance gains across various areas, including reducing code editing errors on the DeepSWE benchmark and significantly improving machine learning research performance on MLE Bench.

Meanwhile, Google is currently testing Gemini 3.5 Pro with partners, and has already begun large-scale pre-training for its next-generation model, Gemini 4.