On December 18, Google officially introduced Gemini 3 Flash - its latest AI model designed to balance expert-level reasoning with lightning-fast performance, all while significantly cutting operational costs. The model is now available for free to users globally.

Tulsee Doshi, Google’s Senior Director of Product Management, presented Gemini 3 Flash as the newest expansion of the Gemini 3 ecosystem. Positioned as a high-efficiency alternative to the premium Gemini 3 Pro, the Flash variant is optimized for real-time responsiveness across a wide range of tasks.

According to Google, the new model retains the core reasoning power of Gemini 3 Pro, yet runs up to three times faster and consumes 30% fewer tokens for everyday tasks, thanks to its dynamic “thinking depth” adjustment based on task complexity.

Performance benchmarks highlight the model’s competitiveness. Gemini 3 Flash scores 90.4% on the GPQA Diamond test (PhD-level knowledge) and 81.2% on the MMMU Pro benchmark, placing it alongside the most powerful models currently on the market.

But the most defining feature is speed. Independent testing by Artificial Analysis shows that Gemini 3 Flash processes tasks three times faster than its predecessor, Gemini 2.5 Pro. The model intelligently varies its processing intensity to save computational resources where possible - without compromising output quality.

Multimodal made easy: AI for everyone

For everyday users, Gemini 3 Flash introduces a more intuitive multimodal experience. One standout feature is "vibe coding" - a system that lets users build full applications or prototypes within minutes, simply by describing their ideas through voice or text. No prior coding experience is needed.

Users can also upload videos or images for real-time computer vision analysis. For instance, the AI can break down a slow-motion golf swing and generate a detailed improvement plan in seconds - just one example of its advanced visual reasoning capabilities.

Gemini 3 Flash now replaces Gemini 2.5 Flash as the default model in the Gemini app. It is also being integrated into Google Search’s new AI Mode.

This integration allows Google Search to handle multi-layered, complex queries with structured answers. Rather than returning a list of links, the system analyzes the question, synthesizes information, and presents a clear, visual response - complete with real-time, location-based updates. This is especially helpful for last-minute travel planning or researching complex academic concepts.

Accessible and developer-friendly pricing

Google is offering highly competitive pricing for developers using Gemini 3 Flash via API. Input tokens cost $0.50 per million, while output tokens are priced at $3 per million.

Developers can access the model through Google AI Studio, Vertex AI, and the new automation platform Google Antigravity.

The launch of Gemini 3 Flash is a calculated move to make next-generation AI accessible to millions while positioning Google as a top contender in both speed and cost in the generative AI space.

Just days earlier, Google’s main rival, OpenAI, released a major update to its image generation tools in ChatGPT. The new "GPT Image 1.5" model offers image rendering speeds four times faster than before.

Notably, OpenAI’s update includes precision editing. Users can now change specific image elements (like shirt color or object placement) without altering the entire composition. A new "Images" tab in the ChatGPT sidebar also offers built-in filters for easier navigation.

The AI race heats up

AI pioneer Geoffrey Hinton, often dubbed the "Godfather of AI," recently told Business Insider that Google is beginning to overtake OpenAI. “The real surprise,” he said, “is that it took Google this long to pull ahead.”

With Gemini 3 Flash, Google signals it’s ready to reclaim its dominance in the generative AI race - not just through power, but through speed, usability, and strategic reach.

Du Lam