Google releases three new Gemini models to cope with fierce competition in the AI field
In the face of increasingly fierce competition in the artificial intelligence market, Three new Gemini series AI models were released on July 21. The three models launched this time each have their own characteristics. As the main model, Gemini 3.6 Flash has improved in programming, multi-modal processing and knowledge work. According to Google, compared with the previous generation Gemini 3.5 Flash, its output token consumption can be reduced by up to 17%, and its usage costs are reduced. The input and output token prices are set at US$1.50 and US$7.50 per million respectively. Gemini 3.5 Flash-Lite focuses on high throughput and low latency tasks. It is suitable for large-scale application scenarios such as agent search and document processing, and provides better performance at a very low cost. It is worth noting that Google has also launched the Gemini 3.5 Flash Cyber model, which is specially optimized for network security scenarios. This model is fine-tuned based on 3.5 Flash and can efficiently discover, verify and repair code security vulnerabilities at a lower cost. Google said that the model will initially be open to government agencies and verified trusted partners through a limited pilot project, aiming to help network security defenders seize the opportunity to fix vulnerabilities before they are exploited, while preventing the model from being abused. The model release comes at a time when Google is facing challenges in the development of its flagship model. The much-anticipated flagship model Gemini 3.5 Pro was originally planned to be launched in June, but it was said that the release time has been delayed for several weeks due to insufficient performance in internal testing, especially in terms of code generation capabilities. Google says the model is currently being tested with partners. At the same time, Google has revealed that it has started pre-training for the next-generation model Gemini 4. In the context of fierce competition in the AI field, Google is responding to pressure from competitors such as Anthropic and OpenAI by emphasizing the cost-effectiveness and efficiency of models. Data shows that in terms of cost, Gemini 3.6 Flash is already lower than similar competing products such as OpenAI’s GPT-5.6 Terra Max.