Google launches Gemini 3 Flash, makes it default model on Gemini app

Google has introduced its newest AI model, Gemini 3 Flash, shortly after the launch of Gemini 3 Pro. This cutting-edge model has quickly become the default option within the Gemini app, enhancing user experience significantly. Free users can engage with the model by asking a limited number of questions directed to its Thinking model, while additional queries are processed by the non-reasoning ‘Fast’ model. According to Google, Gemini 3 Flash exemplifies the balance of speed and intelligence, achieving impressive results on advanced reasoning and knowledge benchmarks such as GPQA Diamond with a score of 90.4% and Humanity’s Last Exam at 33.7% without tools. This performance rivals that of larger frontier models, surpassing the earlier Gemini 2.5 Pro across various evaluations. The company highlighted that Gemini 3 Flash operates three times faster than the Gemini 2.5 Pro on the Artificial Analysis benchmark, all while being more cost-effective. Previously, Gemini 2.5 Pro was among the top-ranked models for months until the introduction of the Gemini 3 Pro last month. In the Humanity's Last Exam benchmark, Gemini 3 Flash scored 33%, outperforming competitors like Grok 4.1 Fast and Claude Sonnet 4.5, while closely trailing behind GPT-5.2. In coding-related tasks evaluated by the SWE-bench Verified benchmark, it achieved a score of 78%, just ahead of Claude 4.5 Sonnet and nearing GPT-5.2. Additionally, Gemini 3 Flash excelled in the ARC-AGI-2 benchmark for visual reasoning, outperforming both Claude and even the latest Gemini 3 Pro model. The best part? Gemini 3 Flash is available to users at no extra cost, now serving as the default model in the Gemini app. It also enhances the AI Mode in Search, allowing users to interact through conversational queries, similar to the Perplexity experience. Stay updated with the latest in technology and more by downloading The Mint News App for daily market updates and business news.