New Delhi: Google has officially introduced its latest Gemini 3.7 Flash as its most intelligent workhorse model yet, specifically aiming at coding, AI agents, and complex knowledge work. Google stated that the model is better at software engineering tasks such as debugging, resolving issues, and generating production-ready code.
It also improves web development and can follow design references more accurately when generating user interfaces. Gemini 3.7 Flash is specially designed to handle multi-step tasks, use tools more effectively, and adapt when it encounters obstacles. This makes it particularly relevant for autonomous AI-agent workflows.
Reports also showed significant improvements over Gemini 3.6 Flash, including 43.6% vs 34.4% on FrontierCode 1.1 Main and 63.5% vs 49.0% on DeepSWE v1.1. Its reported WebDev Arena score increased to 1588, compared with the 1538 for Gemini 3.6 Flash, suggesting stronger performance in generating websites and UI-aimed code. Google also reported gains in complex knowledge tasks, potentially benefiting areas such as finance, law, biosciences and business operations.
The company is offering this latest Gemini 3.7 Flash at an introductory price of $0.75 per million input tokens and $3.75 per million output tokens via December 31, 2026. From January 1, 2027, the announced pricing rises to $1.50 per million input tokens and $7.50 per million output tokens.
That makes the introductory pricing 50% lower than Gemini 3.6 Flash’s original pricing, potentially making it attractive for developers running high-volume AI agents.
Gemini Spark is now powered by the Gemini 3.7 Flash, and it can perform multi-step tasks across Workspace apps such as Gmail, Calendar, and Docs under the user’s direction. Google has also released the Gemini 3.7 Flash only about three weeks after Gemini 3.6 Flash, showing how quickly the company is iterating its AI models.
This debut also comes amid intense competition in AI coding and agentic AI from OpenAI and Anthropic. Google is positioning it as a relatively inexpensive AI workhorse capable of writing code, debugging software, using tools, and executing multi-step business workflows. This makes the model particularly important for the emerging market of autonomous AI agents.
