3 6 Ke·2025-12-18 00:54Core Insights - Google has officially launched Gemini 3 Flash, a lightweight model priced at only 1/5 of Claude and 1/4 of GPT, while outperforming Claude Sonnet 4.5 in coding and competing closely with GPT-5.2 in reasoning and multimodal capabilities [1][7]. Pricing and Performance Comparison - Input price for Gemini 3 Flash is $0.5 per million tokens, while output price is $3 per million tokens, significantly lower than competitors [9][11]. - In various benchmarks, Gemini 3 Flash achieved 90.4% in GPQA scientific reasoning, 81.2% in MMMU-Pro multimodal understanding, and 78% in SWE-bench coding, outperforming Claude Sonnet 4.5 and closely competing with GPT-5.2 [8][10][11]. User Experience and Accessibility - Gemini 3 Flash will replace the previous 2.5 Flash model in the Gemini App, providing all users, including free users, with an upgraded experience without additional costs [13][18]. - The AI Mode in Google Search will also be upgraded to utilize Gemini 3 Flash, enhancing its ability to understand complex user queries and generate clearer responses [15][18]. Developer Benefits - Developers can now utilize Gemini 3 Flash for agentic applications without the high costs associated with flagship models, offering a balance of speed, intelligence, and affordability [19][21]. - Google has introduced cost optimization features such as Context Caching and Batch API to further reduce expenses for high-frequency tasks [19]. Strategic Positioning - The launch of Gemini 3 Flash redefines the "Flash" category, demonstrating that lightweight models can deliver high performance without sacrificing intelligence [20]. - The Gemini 3 family now includes three versions: Gemini 3 Pro for extreme reasoning, Gemini 3 Deep Think for deep analytical tasks, and Gemini 3 Flash for cost-effective, efficient solutions [20].