Red Hot Cyber
Cybersecurity is about sharing. Recognize the risk, combat it, share your experiences, and encourage others to do better than you.
The battle for AI leadership is increasingly moving from the lab to mass production. The day after the release of OpenAI GPT Image 1.5, Google unveiled the Gemini 3 Flash model and began making it the default in its consumer apps, as well as in AI mode in search.
This new model replaces the Gemini 2.5 Flash, released just six months earlier.
Google says Gemini 3 Flash has significantly improved reasoning, multimodality , and efficiency , while maintaining low latency.
In early tests, the model came close to Google and OpenAI’s “frontier” systems. For example, in the Humanity’s Last Exam benchmark, it achieved an uninstrumented score of 33.7%, compared to 37.5% for Gemini 3 Pro, 11% for Gemini 2.5 Flash, and 34.5% for OpenAI’s recently released GPT-5.2 . In the MMMU-Pro multimodal test, Gemini 3 Flash achieved a score of 81.2%, which Google says is the best result among its competitors in this category.
Despite the improved performance, Google positions Flash as a “fast” model. The company claims it outperforms Gemini 2.5 Pro, running about three times faster and consuming about 30% fewer tokens on average in reasoning tasks, potentially reducing implementation costs.
For regular users, Gemini 3 Flash has already become a standard feature of the Gemini app worldwide, while Gemini 3 Pro remains available for more complex math and programming tasks.
Google is also emphasizing multimodal scenarios: uploading short videos for tips and coaching, submitting sketches for recognition, and using audio for analysis or creating quizzes. It’s also noted that the model more often suggests “visual answers,” including images and tables , and helps create simple app prototypes directly within the Gemini interface based on text queries.
In search, Google is simultaneously expanding access to advanced features: Gemini 3 Pro is now available to all users in the United States, and adoption of its proprietary Nano Banana Pro image generation model in search has also increased. In the enterprise space, Google is reporting adoption of Gemini 3 Flash by companies like JetBrains, Figma, Cursor, Harvey, and Latitude, and the model itself is available through Vertex AI andGemini Enterprise .
Developers were promised access to an API preview and integration with Antigravity , a programming tool released by Google last month.
Google has priced the API at $0.50 per million input tokens and $3.00 per million output tokens, slightly more expensive than Gemini 2.5 Flash, but the company attributes this to improved quality.
Google also claims that since the release of the Gemini 3 series, over a trillion tokens are being processed through its API per day . Against this backdrop, competition with OpenAI is intensifying: in early December, several publications reported Sam Altman’s internal “Code Red” memo following a decline in ChatGPT traffic due to Google’s growing share.
OpenAI subsequently released GPT-5.2 and a new image generation model, claiming that ChatGPT message volume had increased eightfold since November 2024.
Follow us on Google News to receive daily updates on cybersecurity. Contact us if you would like to report news, insights or content for publication.
