Hajra Naz
Google DeepMind has introduced three new Gemini AI models. The company announced Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber on Tuesday.
The biggest release is Gemini 3.6 Flash. Google describes it as its new “workhorse” model. It is designed to handle coding, knowledge work, and multimodal tasks more efficiently.
The company says the model uses up to 17% fewer tokens than Gemini 3.5 Flash. That means lower operating costs while maintaining strong performance.
Google also launched Gemini 3.5 Flash-Lite. It is the company’s most affordable model in the Flash family. The model is aimed at developers who need low-cost AI for large-scale applications.
The third release is Gemini 3.5 Flash Cyber. Unlike the other two, this version focuses on cybersecurity. It has been fine-tuned to detect and fix security vulnerabilities.
Google says Flash Cyber offers strong performance at a reasonable cost. However, it won’t be available to everyone. The model will only be offered to governments and trusted partners through a limited pilot program.
According to Google, all three models were built with the same goal. The company wants to improve efficiency, reduce response times, and deliver more reliable performance for businesses building AI agents at scale.
Read More: Google is working on a new AI chip designed to Boost Gemini Performance
The announcement is notable for another reason.
While Google released three new models, it did not unveil the long-awaited Gemini 3.5 Pro. That remains the company’s flagship model for advanced reasoning and coding tasks. The last major update for Gemini Pro arrived in February.
A lot has changed since then.
OpenAI has already released GPT-5.5 and has started rolling out GPT-5.6. Anthropic has also expanded its lineup with Claude Opus 4.8, Claude Sonnet 5, and broader access to its frontier Fable 5 model.
The rapid pace of new releases has increased pressure across the AI industry. Companies are racing to launch more capable models while improving speed and lowering costs.
Google had previously suggested that Gemini 3.5 Pro was close to launch.
When the company introduced Gemini 3.5 Flash in May, it said the Pro model was already being tested internally. At the time, Google expected to release it the following month.
That timeline has changed.
Last week, Bloomberg reported that Google was dealing with internal delays. According to the report, the company has struggled to meet its own performance targets for Gemini 3.5 Pro.
Google’s Pro models are built for more demanding workloads. They are designed to handle advanced reasoning, complex coding, and difficult problem-solving tasks. Flash models, on the other hand, focus on lower costs and faster response times. They are mainly intended for production environments and large-scale AI applications.
Despite the delay, Google says development is continuing.
Google DeepMind Product Lead Logan Kilpatrick said the company is currently testing Gemini 3.5 Pro with selected partners. He added that the team hopes to launch the model soon.
Kilpatrick also revealed that Google has already begun its most ambitious pre-training run yet for Gemini 4, signaling that the next generation of the company’s AI models is already in development.
