Google launches three new models of artificial intelligence Gemini
Google released three new AI models on Tuesday, including Gemini 3.6 Flash, its most powerful, and Gemini 3.5 Flash Cyber, which is tuned for cybersecurity.
Google is trying to compete with rivals such as Anthropic and OpenAI on cutting-edge models and in cybersecurity, an area in which both startups have gained ground this year with powerful systems that can identify security vulnerabilities in software.
Gemini Flash 3.6 improves on the capabilities of Google’s previous version of Flash, which according to AI rankings performed very well in benchmarks such as coding and finance. The new model is also cheaper for users.
The Gemini 3.5 Flash Cyber can find and fix security vulnerabilities at a lower cost than other larger models, the company said. A third model, the Gemini 3.5 Flash-Lite, is designed for tasks such as managing artificial intelligence “agents,” which are robots that can operate autonomously and act as digital personal assistants.
The new Flash models were created to “meet the sweet spot of efficiency and quality,” said Tulsee Doshi, senior director of product management for Google’s Gemini team.
Google is also working on a flagship AI model called Gemini 3.5 Pro, which is expected to be its most powerful system yet. The Silicon Valley company said in May that the Pro would launch in June, but it remains unavailable. Google said the Pro is still in testing and will be “widely available when it’s ready.”
Ever since OpenAI’s ChatGPT chatbot started the AI boom in 2022, Google has invested money and effort in leading the race. Its Gemini AI models soon ranked among the top AI tools.
But in recent months, Google has been boosted by advanced new models from Anthropic and OpenAI, as well as new models from Chinese start-ups like Moonshot AI. Also this month, Meta, the owner of Facebook and Instagram, released a model that surpassed Gemini’s capabilities, according to several charts.
Cybersecurity is increasingly focused on AI models. Systems can find bugs in software and fix them faster than humans, but they can also be used to exploit vulnerabilities and carry out cyber attacks.
In April, Anthropic released a cybersecurity-focused model called Mythos, which it said could create a “cybersecurity showdown.” To prevent the powerful model from falling into the wrong hands, the company has only made it available to a small group of organizations so they can defend against cyber attacks. Anthropic later released the Fable, a model with similar capabilities and a guardrail to prevent it from being used for hacking.
OpenAI soon unveiled its own cybersecurity model, making it available to only a limited group of organizations to prepare their defenses before rolling it out to the wider community. (The New York Times has sued OpenAI and Microsoft, alleging copyright infringement of news content related to AI systems. Both companies have denied the claims.)
Google said it took a similar approach with Gemini 3.5 Flash Cyber. When tested, the model found 55 problems in a complex piece of open-source code known as the V8 JavaScript Engine, including 10 vulnerabilities that no other model had previously detected, the company said.
“The model will be available exclusively to governments and trusted partners,” Ms Doshi said. “This will give front-line defenders a head start on finding and patching critical vulnerabilities before they can be exploited, while mitigating the risk of wider exploitation.”
Google said the new Flash models will be available at a lower price than its previous offerings. The systems use 17 percent fewer tokens, the units of measurement for AI use, than earlier models. This means developers spend less money to complete the same tasks.