In the face of fierce competition in the artificial intelligence market, Google is taking decisive steps to strengthen the position of its Gemini model lineup. On August 16, 2026, the company unveiled a new version — Gemini 3.7 Flash. This release comes against the backdrop of the corporation's efforts to rectify the situation following the controversial launch of previous versions, including Gemini 3.5 Pro. The new model is positioned as a key tool for software code development and the creation of autonomous AI agents.

Acceleration strategy: from 3.6 to 3.7 in three weeks

The speed of updates has become one of the main characteristics of the current stage of Google AI product development. Gemini 3.7 Flash was released just three weeks after the launch of version 3.6 Flash. Such a compressed interval between updates, which would previously have been unthinkable for major technology companies, signals a paradigm shift: Google is moving towards a model of continuous and rapid evolution of its algorithms. Experts link this to the need to regain lost ground and compensate for delays in launching flagship models of the Pro series.

Economic breakthrough: cost reduction with increased power

One of the most significant aspects of the new release was a radical restructuring of pricing. Until the end of 2026, the cost of using Gemini 3.7 Flash will be $0.75 per 1 million input tokens and $3.75 per 1 million output tokens. For comparison, the previous version, Gemini 3.6 Flash, cost $7.50 per million output tokens. Thus, Google has not only increased performance but also halved the product's cost, making it more accessible to developers and corporate clients.

Technical achievements and leadership in benchmarks

The new model demonstrates impressive results in independent tests. In the Agent Arena ranking, which evaluates AI capabilities in performing complex tasks, Gemini 3.7 Flash took a confident 20th place, whereas its predecessor (3.6 Flash) was only in 35th position. In the DeepSWE test, which checks software problem-solving skills, the novelty scored 65.5%. This is 18.8% higher than the result of the previous version. Google states that Gemini 3.7 Flash has become their most intelligent working model, specifically tuned for production code generation.

Ecosystem integration and new features

In addition to improving the model's 'brain', Google has expanded its functional capabilities. According to sources, Gemini 3.7 Flash has received deep integration with Google Workspace, allowing users to work with documents, spreadsheets, and calendars directly within the chat interface. A watermark toggle feature was also added, giving developers more control over AI-generated content. These updates are aimed at making the model not just a text generation tool, but a full-fledged assistant in workflows.