Google reignites the AI battle with 3 new Gemini models

Google is updating its Flash lineup, the one dedicated to lightweight AI, with three new models: Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber. Each of these models has its own specific features. It should be noted that this announcement comes in a particular context for Google.

Google unveils three new AI models

Google has updated its Gemini Flash lineup, meaning its lightweight AI models. First, we find Gemini 3.6 Flash.The company particularly emphasizes the fact that it listened to feedback from developers and customers regarding the 3.5 Flash version to develop this new model. According to Google, it is the « most performant model for everyday use […] », which would offer a significant improvement in coding, better knowledge processing, as well as optimized token usage (these chunks of text analyzed and generated by AI, which also determine usage costs). Indeed, this model would use 17% fewer output tokens than the previous version.

We then find Gemini 3.5 Flash-Lite « designed to scale multi-agent workflows ». More precisely, according to Google, it is the fastest model in the 3.5 series, but also the most economical, reaching 350 output tokens per second.

Finally, the company unveiled Gemini 3.5 Flash Cyber, which, as its name indicates, is designed to detect vulnerabilities in a computer system. Google explains: « Flash’s performance and efficiency make it an ideal foundation for detecting, validating, and fixing security flaws in code at scale ». The Gemini 3.6 Flash and Gemini 3.5 Flash-Lite models are already available.Gemini 3.5 Flash Cyber, meanwhile, is exclusively reserved « for governments and their trusted partners via CodeMender ». Finally, it is worth noting that Google has announced it is working on Gemini 3.5 Pro, but has not given a precise launch date.

A strategic offensive ahead of highly anticipated financial results

Google unveiled its three new artificial intelligence models on the eve of the quarterly results of its parent company Alphabet. Of course, this timing is no coincidence. The company is seeking to reassure investors while reaffirming its position in the race to develop AI. That frenzious race is currently dominated by Anthropic, which regularly announces lightning-fast advances, and OpenAI, which notably claims to have lost control of two AI systems because of how capable they are.

On its side, Google appears to be somewhat behind. Gemini 3.5 Pro is taking its time, and it cannot be said that the current models offer lightning-fast performance. At least for now. On X, Logan Kilpatrick, product lead at Google AI Studio, explained: « We have launched the most ambitious pretraining session to date for Gemini 4, and we are very excited about the progress made ». It now remains to be seen whether Gemini 4 can compete with the leading players in the sector.

👉🏻 Follow tech news in real time: add 01net to your sources on Google, and subscribe to our WhatsApp channel.

Source: Google