Google Rolls Out Gemini 3 Flash Globally
On December 17, Google announced “Gemini 3 Flash,” a new AI model focused on efficiency. While maintaining the advanced reasoning capabilities of Gemini 3 Pro (released November 18), it achieves significantly faster speeds and lower costs.
Outperforming Gemini 2.5 Pro
According to Google, Gemini 3 Flash outperforms the previous flagship model, Gemini 2.5 Pro, while operating at 3x the speed. This results in dramatically improved cost efficiency.
Comparison with GPT-5.2
In benchmark tests, Gemini 3 Flash shows comparable results to OpenAI’s GPT-5.2. Notable findings include:
- Humanity’s Last Exam: Less than 1 percentage point difference from GPT-5.2 without tool access
- Multiple benchmarks: Outperforms GPT-5.2 in several tests
Now the Default Model in Gemini App
Gemini 3 Flash is rolling out globally as the default model in the Gemini app (web and mobile), replacing Gemini 2.5 Flash.
It’s also becoming the default model for AI Mode in Google Search, available alongside Gemini 3 Pro and Nano Banana Pro.
Key Features
Multimodal Reasoning
Capable of analyzing video, images, audio, and text to generate content. For example, it can create plans from multiple videos and images in just seconds.
Developer Availability
- Available in Google AI Studio and Gemini CLI
- Enterprise access through Vertex AI and Gemini Enterprise
- Already adopted by JetBrains, Figma, Cursor, Harvey, and Latitude
OpenAI’s “Code Red” Response
The release of Gemini 3 Pro reportedly triggered a “code red” at OpenAI, leading to the emergency release of GPT-5.2. This Flash model is likely to further accelerate the AI supremacy race.