Google Unveils Gemini 3 Flash, Sets It as Default in Gemini App

Google has launched its cost-effective and speedy Gemini 3 Flash model, building on the Gemini 3 foundation released last month. This move aims to challenge OpenAI's momentum. Google is also setting this model as the default within the Gemini app and its AI-powered search.

This new Flash model arrives six months after the Gemini 2.5 Flash, delivering substantial advancements. In benchmark tests, Gemini 3 Flash significantly outperforms its predecessor and matches the capabilities of other leading models like Gemini 3 Pro and GPT-5.2 in several key areas.

For example, on the Humanity's Last Exam benchmark—designed to test cross-domain expertise—it achieved a score of 33.7% without using external tools. Comparatively, Gemini 3 Pro scored 37.5%, Gemini 2.5 Flash scored 11%, and the recently released GPT-5.2 scored 34.5%.

On the multimodal reasoning benchmark MMMU-Pro, the new model led all competitors with a score of 81.2%.

Consumer rollout

Google is making Gemini 3 Flash the default model in the Gemini app worldwide, replacing Gemini 2.5 Flash. Users can still select the Pro model from the options for tasks involving math and coding.

The company states the new model excels at interpreting multimodal content to provide relevant answers. You could, for instance, upload a short pickleball video and request tips, draw a sketch for the model to identify, or submit an audio recording for analysis or quiz generation.

Google also highlighted the model's improved understanding of user query intent and its ability to generate more visually rich responses, incorporating elements like images and tables.

Join the Disrupt 2026 Waitlist

Add yourself to the Disrupt 2026 waitlist for priority access when Early Bird tickets are released. Past Disrupt events have featured industry leaders from Google Cloud, Netflix, Microsoft, Box, Phia, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, and Vinod Khosla—part of over 250 experts leading 200+ sessions designed to accelerate your growth. Explore innovations from hundreds of startups across all sectors.

Join the Disrupt 2026 Waitlist

Add yourself to the Disrupt 2026 waitlist for priority access when Early Bird tickets are released. Past Disrupt events have featured industry leaders from Google Cloud, Netflix, Microsoft, Box, Phia, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, and Vinod Khosla—part of over 250 experts leading 200+ sessions designed to accelerate your growth. Explore innovations from hundreds of startups across all sectors.

The new model can also be used to create app prototypes directly within the Gemini app using simple prompts.

Gemini 3 Pro is now available to all users in the U.S. for search, and more people in the U.S. can also access the Nano Banana Pro image model in search results.

Enterprise and developer availability

Google noted that companies including JetBrains, Figma, Cursor, Harvey, and Latitude are already utilizing the Gemini 3 Flash model, accessible via Vertex AI and Gemini Enterprise.

For developers, the model is available in a preview state through its API and within Antigravity, Google's recently launched coding tool.

Google reported that Gemini 3 Pro scores 78% on the SWE-bench verified coding benchmark, a figure only surpassed by GPT-5.2. The model is described as ideal for video analysis, data extraction, and visual Q&A, and its speed makes it well-suited for rapid, repetitive workflows.

Pricing for the model is set at $0.50 per 1 million input tokens and $3.00 per 1 million output tokens. This represents a slight increase over Gemini Flash 2.5's rates of $0.30/$2.50. However, Google asserts that the new model surpasses the performance of Gemini 2.5 Pro while operating three times faster. For complex reasoning tasks, it also uses an average of 30% fewer tokens than the 2.5 Pro model, potentially lowering overall token costs for specific applications.

"We position Flash as more of a workhorse model. Looking at the input and output pricing at the top of this table, Flash is a significantly more economical option. This makes it practical for companies to handle bulk tasks," explained Tulsee Doshi, senior director & head of Product for Gemini Models, in a TechCrunch briefing.

Since launching Gemini 3, Google has processed over one trillion tokens daily through its API, reflecting the intense competitive race with OpenAI on releases and performance.

Earlier this month, reports indicated Sam Altman sent an internal "Code Red" memo at OpenAI following a dip in ChatGPT traffic as Google's consumer market share grew. In response, OpenAI released GPT-5.2 and a new image generation model, also touting expanded enterprise adoption and an eightfold increase in ChatGPT message volume since November 2024.

While not directly addressing OpenAI, Google commented that the pace of new model releases is pushing all companies in the industry to stay highly active.

"What's happening across the industry is that all these models continue to impress, challenge one another, and push the boundaries. It's also exciting to see how companies are deploying them," Doshi said.

"We're also introducing new benchmarks and evaluation methods for these models, which is further driving innovation."