Google’s Gemini

Google has announced another major step in advancing artificial intelligence (AI), unveiling its Gemini 3.7 Flash model with a focus on developing more capable AI agents.

The new model is designed primarily as a “workhorse” AI system, particularly for coding, software engineering, web development and other complex tasks. Google said Gemini 3.7 Flash can deliver significantly improved performance compared with its predecessor, Gemini 3.6 Flash.

A new approach to performance: One of the key improvements in Gemini 3.7 Flash is its ability to break down complex tasks into multiple steps and complete them systematically.

The model is expected to be more reliable in identifying problems in software, debugging them and generating appropriate code. It can also better understand instructions and adjust its approach if it encounters problems while completing a task.

The model is also expected to perform strongly in web development. It can analyse an image, design or screenshot and create a website with features that closely resemble the reference.

Built for autonomous tasks: Unlike conventional chatbots, which mainly respond to user queries, AI agents can plan multiple steps to accomplish a task and use appropriate tools along the way. Gemini 3.7 Flash has been specifically developed to perform such agentic tasks.

Google said the model can understand what a task requires and determine the next steps on its own.

The model has also been incorporated into Google’s Spark version, where it can use several Google Workspace tools to perform tasks such as drafting emails and updating status documents.

Access and pricing: The cost of using Gemini 3.7 Flash will depend on the level of usage. Developers and other interested users can access the model through the Gemini API, Google AI Studio, Android Studio and Google Antigravity.

Google is expected to continue investing in research and development to further improve the capabilities of its AI models and take AI-powered agents to a new level.