Google's Gemini 3 Flash model outperforms GPT-5.2 in some benchmarks (www.engadget.com)

Google has launched the Gemini 3 Flash model, an efficient spin-off of its Gemini 3 Pro, which promises high-level reasoning capabilities at a significantly reduced cost. This rollout comes just a month after the Pro version debuted. The Flash model has shown impressive benchmark results, outperforming previous iterations like Gemini 2.5 Pro and rivaling OpenAI's GPT-5.2. Notably, in the challenging Humanity's Last Exam suite, it scored nearly on par with GPT-5.2, and in the MMMU-Pro benchmark focused on multimodal reasoning, it surpassed OpenAI's offering with a score of 81.2 percent compared to 79.5 percent for GPT-5.2. The implications of this development are significant for the AI/ML community, as it indicates a growing competitiveness in generative AI, particularly between Google and OpenAI. With access to the Gemini 3 Flash model incorporated into Google’s Gemini App and AI Mode in Search at no additional cost to users, Google aims to enhance everyday tasks for billions of users. This launch not only positions Google as a formidable player in the AI space but also puts pressure on OpenAI to further innovate, potentially altering the landscape of AI applications and accessibility.