Gemini 3 Flash: Google’s New AI Model Now Available to Millions Worldwide

Google is stepping on the gas again. The company unveiled Gemini 3 Flash on Wednesday, a new AI model designed to deliver powerful reasoning at much faster speeds and at a lower cost.

Gemini 3 Flash is the newest member of Google’s Gemini 3 model family. It is built to be quick, efficient, and cheaper to run, while still keeping the advanced reasoning found in larger models.

In a blog post announcing the release, Tulsee Doshi, senior director of Product Management for Gemini models, wrote that Gemini 3 Flash is designed to bring advanced intelligence to more people, faster. According to Google, the model combines “Pro-grade” reasoning with low latency. This means it can respond quickly without sacrificing quality.

The release follows last month’s launch of Gemini 3 Pro and Gemini 3 Deep Think, which Google says have already been handling massive demand. Since then, the company says it has been processing more than one trillion tokens per day through its API.

Now the default for millions of users

Google is not limiting Gemini 3 Flash to developers. The company says the model is now rolling out globally and becoming the default option in several major places.

Gemini 3 Flash is replacing Gemini 2.5 Flash as the default model in the Gemini app, meaning users around the world will get access to the new system at no cost. It is also becoming the default model for AI Mode in Google Search.

Robby Stein, vice president of Product for Google Search, said in a blog post that Gemini 3 Flash brings advanced reasoning “at the speed you expect of Search.” He added that AI Mode is now better at understanding complex questions and returning detailed, well-formatted answers while still pulling real-time information from the web.

Strong performance, lower cost

Google says Gemini 3 Flash delivers what it calls “frontier intelligence built for speed.” Despite being lighter and cheaper to run, the model performs strongly on tough benchmarks.

According to Google’s disclosures, Gemini 3 Flash scored 90.4% on GPQA Diamond, a PhD-level reasoning benchmark, and 33.7% on Humanity’s Last Exam without tools. On the multimodal MMMU Pro benchmark, it posted an 81.2% score, matching or outperforming several larger models.

The company also says the model is more efficient in day-to-day use. Gemini 3 Flash reportedly uses 30% fewer tokens on average than Gemini 2.5 Pro for common tasks, while running up to three times faster.

Pricing reflects that focus on scale. Gemini 3 Flash costs $0.50 per 1 million input tokens and $3 per 1 million output tokens, positioning it as one of Google’s more affordable advanced models.

On the SWE-bench Verified benchmark, Gemini 3 Flash achieved a 78% score, outperforming Google’s earlier 2.5 models and even surpassing Gemini 3 Pro in that test.

The model is now available to developers through the Gemini API in Google AI Studio, Gemini CLI, and Google’s new agent-building platform, Antigravity. Enterprise customers can access it via Vertex AI and Gemini Enterprise.

The AI race heats up

The launch of Gemini 3 Flash comes amid intense competition with OpenAI and other AI labs. Google’s release of Gemini 3 Pro in November reportedly triggered an internal alarm at OpenAI. In the following weeks, OpenAI fired back with its own GPT-5.2 model.

With Gemini 3 Flash, Google is leveraging its massive advantage: ubiquity. By baking it directly into search and its free app, it’s putting this new tech in front of billions overnight, hoping its blend of speed and intelligence will be the new standard.

Want more on Google’s AI push? Check out how the company is also rolling out a CC AI agent, signaling an even bigger bet on agent-driven automation and enterprise AI.