In response to the increasingly intense competition in the artificial intelligence market, Alphabet Inc. (NASDAQ: GOOG) released three new models in its Gemini AI series on July 21st.
Overview of the New Models
The three newly launched models each have distinct characteristics. Gemini 3.6 Flash serves as the main workhorse model, showing improvements in programming, multimodal processing, and knowledge work. According to Alphabet, compared to its predecessor Gemini 3.5 Flash, it can reduce output token consumption by up to 17% while lowering usage costs. The pricing for input and output tokens is set at $1.50 and $7.50 per million, respectively. Gemini 3.5 Flash-Lite is designed for high-throughput, low-latency tasks, making it suitable for scalable applications like agent search and document processing, offering superior performance at a very low cost.
Specialized Security Model
Notably, the company also introduced the Gemini 3.5 Flash Cyber model, specifically optimized for cybersecurity scenarios. This model is fine-tuned from the 3.5 Flash base and can efficiently discover, verify, and fix code security vulnerabilities at a lower cost. The company stated that this model will initially be available only to government agencies and verified, trusted partners through a limited pilot program. The aim is to help cybersecurity defenders gain an advantage by fixing vulnerabilities before they can be exploited, while also preventing potential misuse of the model.
Context of the Launch
This model release coincides with challenges the company is facing in developing its flagship model. The highly anticipated flagship model, Gemini 3.5 Pro, was originally scheduled for a June launch. However, it is reported that the release has been delayed by several weeks because it failed to meet internal testing targets, particularly in code generation capabilities. Alphabet indicated the model is currently being tested with partners. Concurrently, the company has revealed it has begun pre-training work on its next-generation model, Gemini 4.
Competitive Landscape
Amid the white-hot competition in the AI field, Alphabet is responding to pressure from competitors like Anthropic and OpenAI by emphasizing the cost-effectiveness and efficiency of its models. Data shows that in terms of cost, Gemini 3.6 Flash is already priced lower than comparable competing products such as OpenAI's GPT-5.6 Terra Max.
Comments