Google launches Gemini 3.5 Flash and the Gemini Spark agent: what changes

The Google I/O 2026 made it clear that the AI race is fiercer than ever. The company announced two highlights set to shake the market: Gemini 3.5 Flash, a lighter and cheaper version of its AI model, and Gemini Spark, a general-purpose agent able to reason across connected apps. The message to the industry is direct: Google does not intend to trail OpenAI and Anthropic.

What is Gemini 3.5 Flash

Gemini 3.5 Flash is Google bet on something that became an industry obsession: delivering cutting-edge capability at a far lower cost. According to the company, the model offers performance comparable to frontier models at half — in some cases a third — of the price of equivalent rivals. In practice, developers and companies can run more sophisticated AI applications while spending much less.

This “light but powerful” strategy is not new, but Google is betting big. Fast, cheap models are ideal for high-volume tasks: customer support, automatic summaries, content classification and assistants embedded in products. The lower the cost per request, the more viable it becomes to integrate AI at scale.

Gemini Spark: the agent that thinks across your apps

If Flash is about efficiency, Gemini Spark is about ambition. It is a general-purpose AI agent inside the Gemini app, able to reason over information spread across connected apps. Instead of opening each service manually, the agent cross-references data from different sources to carry out more complex tasks.

For now, Spark is in beta and will be released first to “trusted testers” and Google AI Ultra subscribers. It is Google bet on the hottest trend of 2026: AI agents, systems that do not just answer questions but act on the user behalf.

Why this matters to you

Pairing a cheap model (Flash) with a capable agent (Spark) shows where the market is heading. Instead of passive chatbots, we are moving toward assistants that complete tasks end to end: organizing your schedule, researching and comparing products, drafting and sending messages. For professionals and small businesses, that can mean real productivity gains.

The competitive context

The announcement does not happen in a vacuum. OpenAI shipped GPT-5.5 just weeks earlier, and Anthropic has kept an aggressive pace of Claude updates. Competitive pressure is pushing prices down and quality up — a direct benefit to end users and developers.

For Google, which historically led AI research but saw OpenAI steal the spotlight with ChatGPT, I/O 2026 is a show of strength. The message is that the company has the technical muscle and infrastructure to compete on the two fronts that matter: cost and capability.

What to expect next

The expansion of Gemini Spark to more users will be the thermometer. If the agent delivers, we may see a shift in how we interact with software day to day. Gemini 3.5 Flash, in turn, should accelerate AI adoption in products that previously could not afford premium models.

At DigitalRadar, we will follow closely how these launches translate into practical features. One thing is certain: 2026 is the year AI stopped being a promise and became an everyday tool.

Conclusion

With Gemini 3.5 Flash and Gemini Spark, Google reinforces that the next AI frontier is not only about the smartest model, but the most useful and affordable one. For those building products or simply enjoying the technology, this is great news.