Google has announced three new Gemini models designed for developers who want to build AI agents at scale. Gemini 3.6 Flash and 3.5 Flash-Lite are available immediately. Gemini 3.5 Flash Cyber, in turn, is designed to detect and address vulnerabilities.
The new models are designed to bridge the gap between efficiency and quality, which is precisely the purpose of Google’s Flash series. Gemini 3.6 Flash serves as the workhorse for coding, knowledge work, and multimodal tasks. According to Google, the model consumes 17 percent fewer output tokens than its predecessor, 3.5 Flash, while the cost per token is actually lower: $1.50 per million input tokens and $7.50 per million output tokens.
Google is also introducing Gemini 3.5 Flash-Lite. This is the fastest model in the 3.5 series, with a throughput of 350 output tokens per second. It is designed for high-volume, low-latency tasks, such as agentic search and document processing. The price is $0.30 per million input tokens and $2.50 per million output tokens.
Both models now include Computer Use as a built-in tool, enabling agents to interact independently with interfaces and websites. They also support a context window of up to 1 million tokens.
Third model exclusively for governments
The third new model, Gemini 3.5 Flash Cyber, is a different story. This is a specialized cybersecurity model built on top of 3.5 Flash and trained to detect and remediate vulnerabilities. It runs within Google’s CodeMender agent.
Due to its dual-use nature, Flash Cyber will not be rolled out widely. Google is making it available exclusively to governments and trusted partners as part of a limited-access pilot program. In this way, the company aims to give defenders a head start and reduce the risk of misuse.
Gemini 3.5 Pro is still a work in progress
The notable absentee is Gemini 3.5 Pro. Google is currently testing that model with partners and plans to release it widely only once it is ready. According to Bloomberg reports, 3.5 Pro is months behind schedule. At the same time, Google confirms that it has begun pre-training Gemini 4.
Gemini 3.6 Flash and 3.5 Flash-Lite are available as of today via the Gemini API, Google AI Studio, and Android Studio. For businesses, they’ll be coming to the Gemini Enterprise Agent Platform. 3.5 Flash-Lite will also appear in Google Search.
Tip: Panic over Chinese AI model strikes again, and once again, it’s overblown