Google's AI Evolution: From Gemini 3.5 Flash to 3.6 Flash and Beyond
Google's AI journey continues to captivate and intrigue, with the latest developments at I/O in May and the recent announcement of three new AI models. The company's focus on efficiency and user feedback has led to significant improvements in its Gemini models, particularly in coding and multimodal capabilities. However, the delay in the launch of Gemini 3.5 Pro and the deprecation of 3.5 Flash have raised questions about Google's strategy and the future of its AI offerings.
One thing that immediately stands out is Google's commitment to innovation and its willingness to adapt based on user feedback. The company's response to the shortcomings of Gemini 3.5 Flash in code generation has resulted in the creation of Gemini 3.6 Flash, which promises to be marginally more capable and better at coding. This is particularly interesting given the intense focus on efficiency as businesses grapple with the cost of AI tokens. In my opinion, this highlights Google's ability to balance innovation with practical considerations, ensuring that its AI models remain accessible and cost-effective for a wide range of users.
What many people don't realize is that the improvements in Gemini 3.6 Flash are not just incremental. The model's ability to complete tasks more accurately, in fewer steps, and with fewer tokens is a significant advancement. This is especially true in agentic workflows, where the new model's lower API cost of $1.50/1M input tokens and $7.50/1M output tokens could save developers and Google a lot of money. The fact that Gemini 3.6 Flash uses about 17 percent fewer tokens is a testament to Google's commitment to efficiency and its ability to deliver on its promises.
However, Google is not done with the 3.5 branch yet. The release of Gemini 3.5 Flash Lite and 3.5 Flash Cyber further demonstrates the company's commitment to innovation and its willingness to explore new use cases. Gemini 3.5 Flash Lite, with its impressive 350 tokens per second, is ideal for scaling agentic systems without breaking the bank. This is particularly fascinating given the benchmark numbers, which show that the new Flash Lite is almost on par with frontier models from about a year ago, but at a much lower cost. This raises a deeper question: How will the market respond to these new offerings, and what does it mean for the future of AI development and deployment?
From my perspective, Google's AI evolution is a testament to the company's ability to innovate and adapt. The improvements in Gemini 3.6 Flash and the release of Gemini 3.5 Flash Lite and 3.5 Flash Cyber demonstrate the company's commitment to delivering on its promises and exploring new use cases. However, the delay in the launch of Gemini 3.5 Pro and the deprecation of 3.5 Flash raise questions about the company's strategy and the future of its AI offerings. One thing is certain: Google's AI journey is far from over, and the company's commitment to innovation and user feedback will continue to shape the future of AI development and deployment.
