The speed record in the field of artificial intelligence has been broken again. At today's Google I/O 2026 developer conference, Google officially launched the new large model Gemini 3.5 Flash. As the most powerful agent and programming model in Google's history, this model not only approaches the intelligence level of large flagship models but also continues to maintain the characteristic high-speed response advantage of the Flash series.
What excites the industry even more is that Google announced the model is now available to all users worldwide and can be used completely free of charge. This move undoubtedly further lowers the application threshold of top AI technology, offering developers a more cost-effective computing power option.
Outperforms Previous Flagship in Multiple Benchmark Tests
In the official evaluation data, Gemini 3.5 Flash performed impressively in multiple benchmark tests targeting agents and coding, even surpassing the previous flagship model Gemini 3.1 Pro. In particular, in the multimodal understanding test CharXiv Reasoning, it achieved the highest score of all.
Aside from the leap in intelligence, the new model also demonstrated impressive performance in operational efficiency. Data shows that Gemini 3.5 Flash's output speed has exceeded 280 tokens per second, which is four times faster than GPT-5.5 and Claude Opus 4.7.
Significantly Reduces Costs for Long-Term Tasks
According to Google's official statement, due to the excellent balance between speed and performance, Gemini 3.5 Flash is especially suitable for handling long-term complex agent tasks. In practical work scenarios, it can significantly shorten the development time for relevant personnel.
