MUMBAI: Google is giving its AI a faster gear. The company has launched Gemini 3.7 Flash, a new model aimed at software development, web applications and automated business workflows, positioning it as a faster and lower-cost option for developers building AI agents.
The launch comes just three weeks after Gemini 3.6 Flash and focuses on tasks involving multi-step reasoning, tool use and accuracy. Google said the new model can improve debugging, issue resolution and first-pass code generation while reducing the manual intervention needed for complex workflows.
The jump is visible in the software benchmarks cited by Google. Gemini 3.7 Flash scored 43.6 per cent on FrontierCode 1.1 Main, compared with 34.4 per cent for Gemini 3.6 Flash. On DeepSWE v1.1, it scored 65.3 per cent, up from 49 per cent.
Web development has also received a boost. Google said the model can build more complete applications and functional layouts with fewer iterations, while improving its ability to reproduce designs from screenshots, images and design systems. On WebDev Arena, Gemini 3.7 Flash posted an Elo score of 1,588, compared with 1,538 for its predecessor.
The model also recorded higher scores on knowledge-heavy and automation tasks. On the GDP.pdf benchmark, it scored 34 per cent, against 22 per cent for Gemini 3.6 Flash, while its AutomationBench score reached 30.4 per cent, compared with 17 per cent previously.
Google said Gemini 3.7 Flash is designed to spend more effort on planning and tool calls, deal with obstacles, clarify requirements when needed and follow instructions more closely. The aim is to cut down retries and reduce the amount of human oversight required in engineering workflows.
Lower pricing
Google is also putting a lower price tag on the model at launch. Through the end of 2026, Gemini 3.7 Flash will be offered at an introductory rate of $0.75 per million input tokens and $3.75 per million output tokens.
Google said the introductory pricing is 50 per cent lower than the original per-token pricing of Gemini 3.6 Flash.
The model is available through the Gemini API, Google Antigravity, Google AI Studio and Android Studio, while enterprises can access it through the Gemini Enterprise Agent Platform and Gemini Enterprise app.
Gemini 3.7 Flash is also being added to Gemini Spark, Google’s personal AI agent available to Google AI Pro and Ultra subscribers in more than 160 countries. Google expects the upgrade to improve Spark’s ability to handle Google Workspace tasks such as consolidating files, drafting emails and updating status documents.
Google has demonstrated the model across applications including playable 3D games, interactive landing pages, robotics training and converting static annual reports into interactive web experiences.
The company said the model also includes updated safeguards covering potential misuse involving chemical, biological, radiological and nuclear risks, as well as cyber offences.
The launch comes as Google continues work on its next premium Gemini model, keeping the race for increasingly capable AI models firmly in motion.