Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber (blog.google)

Article summary

Google has introduced new Gemini models, including 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber, which aim to deliver improved efficiency, latency, and reliability for building AI agents at scale. The 3.6 Flash model is said to be 17% more token-efficient than its predecessor and performs better in coding and tool usage benchmarks. The models are designed to meet the needs of developers and customers building production AI agents. The company is also working on the next generation of models, including Gemini 4.

Main themes

• AI model updates
• Gemini models
• Efficiency and performance
• AI development
• Google innovations
• Model comparisons

What commenters say

• The new Gemini models may not be impressive compared to other models available in the market, and the benchmarks provided are not convincing.
• The 3.6 Flash model is more expensive than some other models, such as GLM 5.2, but its performance is not significantly better.
• The Gemini models have an advantage in multimodal tasks, such as image and audio analysis, but may not be the best choice for other tasks.
• The lack of comparison to other models in the article is a significant omission, making it difficult to evaluate the true capabilities of the new Gemini models.
• The cost and performance of the Gemini models are not competitive with other models, and the company's focus on Intelligence as a Service may not be enough to justify the expense.
• The Gemini models have shown promising results in certain benchmarks, but more testing and comparison are needed to fully understand their capabilities.
• The development of the Gemini models is a step in the right direction, but the company needs to do more to catch up with other leaders in the field.
• The article's focus on the Gemini models' efficiency and performance may be misleading, as other factors such as cost and compatibility are also important considerations.