Google has now formally released Gemini 3 Flash—which, in the Gemini app globally, has swiftly become the default model. This transition brings an end to the widely-trusted previous model, Gemini 2.5 Flash, which was released just six months ago. The rollout first targets the user3 experience, resulting in important performance and cost efficiency improvements.
Gemini 3 Flash represents an excessive leap in progress compared to the previous version. To enhance speed in task execution and token consumption. This update is further representation of how Google has been positioning itself to compete with industry giants like OpenAI. It cements Google’s dominance in the AI race.
Key Features and Pricing
Gemini 3 Flash is extremely competitive, price-wise, especially compared to its competition. Input tokens cost $0.50 per million purchased. Output tokens cost $3.00 per million purchased. By contrast, Gemini 2.5 Flash cost $0.30 and $2.50, respectively. This newly adopted pricing is a move by Google to show that they want to be a cheaper option for users who are running larger tasks.
Our model uses an average of about 30% fewer tokens for the tasks that require more reasoning than Gemini 2.5 Pro. Gemini 3 Flash delivers phenomenal speeds, processing information up to three times quicker than Gemini 2.5 Pro. These attributes make it an appealing option for developers and businesses looking to increase efficiency without sacrificing quality.
Gemini 3 Flash has shown very strong performance on the most recent benchmarks. It set a new state-against-state record with 81.2% precision on multimodality and reasoning benchmark, MMMU-Pro. This achievement underlines its deep reasoning skills, far-reaching adaptability and versatility to approach various tasks.
User-Friendly Innovations
Gemini 3 Flash is unparalleled in its ability to assist users in crafting app prototypes. You can do this in a few clicks within the Gemini app, or even in just a few words through prompts. This improved functionality encourages a new, more collaborative and immersive approach to development. It opens the door for users to quickly visualize their ideas and deploy them at scale.
Nobody appreciates competition more than Tulsee Doshi, senior director and head of Product for Gemini Models. She’s convinced it’s the primary driver for innovation in the sector. He stated,
“Just about what’s happening across the industry is like all of these models are continuing to be awesome, challenge each other, push the frontier.”
This phrase stands out because it shows that Google is serious about strengthening its offerings and reacting to the changing market landscape.
Expanding Access and Capabilities
Since Gemini 3’s launch, Google has been claiming to process more than 1 trillion tokens per day on its API. This increase is a testament to the overwhelming demand we’ve seen for its AI services. Picture this … and other things to know here at the Smart City lab Gemini 3 Pro is now available to all users in the U.S., furthering Google’s advanced AI capabilities. More users nationwide will be able to use the Nano Banana Pro image model in image-based search functionalities.
Doshi further underscored the company’s goal of advancing evaluation metrics for AI models, ensuring,
“We’re also introducing new benchmarks and new ways of evaluating these models. And so that’s also encouraging us.”
These efforts illustrate Google’s dedication to not only improving its products but advancing how AI models are assessed and compared within the industry.