Google has launched its latest AI model, Gemini 3.7 Flash, aimed at improving coding, software engineering, and knowledge work for developers and enterprise users.
The model, which follows the recent release of Gemini 3.6 Flash, is claimed to offer advancements in several technical benchmarks. It is available immediately at an introductory price set at half of the 3.6 Flash model’s original rate per million tokens.
Go deeper with GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
Gemini 3.7 Flash is said to have demonstrated higher first-pass accuracy and better performance in tasks such as debugging, issue resolution, and production-ready code generation.
Measured against FrontierCode 1.1 Main, Gemini 3.7 Flash achieved a 43.6% score compared to the previous version’s 34.4%. On DeepSWE v1.1, it recorded 65.3%, up from 49.0% with 3.6 Flash.
Web development capabilities have also been updated. The new model reportedly creates more functional layouts and completes apps with fewer prompts.
On Arena.ai’s WebDev Arena, Gemini 3.7 Flash received an Elo score of 1588, exceeding 3.6 Flash’s score of 1538.
Google claims that the model performs well in generating user interfaces from various reference designs including screenshots and images.
In more complex fields such as finance, law, and biosciences, Gemini 3.7 Flash showed increased accuracy and reasoning, according to the company blog by Tulsee Doshi, who serves as Google DeepMind’s senior director and head of product for Gemini models.
The model outperformed 3.6 Flash on the GDP.pdf benchmark, reaching 34.0% against 22.0%. It also produced stronger results on AutomationBench, a test for real-world business workflow automation, by posting a score of 30.4% compared to 17.0%.
Gemini 3.7 Flash is being offered at a temporary rate of $0.75 per one million input tokens and $3.75 per one million output tokens through the end of the year.
The AI model will also update capabilities in Gemini Spark, Google’s AI agent product available in more than 160 countries for AI Pro and Ultra subscribers.
The company says Spark will see improvements in output quality and tool use for Google Workspace apps.
Google has stated that the new model includes updated safeguards to address concerns in chemical, radiological, biological, and nuclear misuse, as well as cyber offence.
Gemini 3.7 Flash can be accessed via the Gemini API, Google AI Studio, and Android Studio for developers, and is available on Google’s enterprise and consumer agent platforms for business and individual users.
In July, Accenture Edge, part of Accenture, and Google Cloud launched a set of agentic AI solutions designed to support mid-market firms with technology transformation.
