Follow ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways
- Google unveiled Gemini 3 Flash on Wednesday.
- It’s the default model in AI Mode in Search and the Gemini app.
- It outperformed Gemini 3 Pro on agentic coding.
Google has unveiled Gemini 3 Flash, the company’s latest AI model, which it says balances speed with industry-leading reasoning and multimodal capabilities.
Along with its global release, Gemini 3 Flash has also been made the default model in both Google Search, the Gemini app, and AI Mode. That’s significant, since in the earlier days of the AI race, companies — even the more powerful ones like Google — would tend to roll out new AI tools slowly and cautiously, getting a sense of user feedback before adding them to their most popular and widely used services.
Also: Inside the making of Gemini 3 – how Google’s slow and steady approach won the AI race (for now)
Last month, Google embedded the recently arrived Gemini 3 into Search; then, as now, the decision to move quickly and upgrade the company’s cornerstone products with a brand new model points strongly to a strong sense of confidence within the company as it steps up its competition with OpenAI and other developers.
“For too long, AI forced a choice: big models that were slow and expensive, or high-speed models that were less capable,” Josh Woodward, VP of Google Labs and Gemini, wrote in a blog post published Wednesday. “Gemini 3 Flash ends this compromise. Gemini 3 Flash delivers smarts and speed.”
Practical implications
At this point in the AI race, you might be feeling a little motion sickness from the sheer number of new models and tools that have been dropping pretty much on a daily basis. Companies have an incentive to move quickly and promote every new AI product as a complete game-changer, since it adds to their aura of power.
It’s therefore always important to ask: What does this or that tool actually offer? Is it just the same old AI capabilities in a new guise, or is there something that’s actually, substantively new, here?
If you write code for a living (or just for fun), you’re likely to see real benefits from Gemini 3 Flash. The model combines Gemini 3 Pro’s frontier coding skill with low latency, and it also offers state-of-the-art agentic capabilities, “which means it can enable more intelligent applications — like live customer support agents or in-game assistants — that demand both quick answers and deep reasoning,” Tulsee Doshi, a senior director of product management at Google, wrote in a blog post.
Also: Google just rolled out Gemini 3 to Search – here’s what it can do and how to try it
For the rest of us, the global deployment of Gemini 3 Flash will primarily mean a boost in the perks they’ve already come to expect from Gemini app or AI Mode in Search.
For example, non-coding experts can feed a stream of consciousness description about an app they’d like to build into the app, and Gemini will quickly spit out a prototype. Or someone running a search can ask for travel advice and reservations for a last-minute vacation, and the model will pull up results based on that user’s particular preferences.
Users will also probably notice a significant boost in the quality and flexibility of the multimodal capabilities offered by Gemini through AI Mode and the app: it can respond to questions about images, video, audio, or text, and thanks to Nano Banana Pro, generate images in AI Mode (just select “Thinking with 3 Pro” and then “Create Images Pro” in the model drop-down menu).
The model fell just short of matching GPT-5.2’s performance on Humanity’s Last Exam, and it outperformed both Gemini 3 Pro and Claude Sonnet 4.5 on the agentic coding SWE-Bench Verified benchmark, according to data published by Google.
Gemini 3 Flash is available now in preview through the Gemini API in Google AI Studio, and it’s starting to roll out on Wednesday for free to all users in the Gemini app and in AI Mode in Search.
