Google has launched Gemini 3.7 Flash, its latest artificial intelligence model designed for coding, agentic workflows and enterprise tasks, as the company looks to improve the performance and economics of AI systems used at scale.
The model arrives three weeks after Gemini 3.6 Flash and brings improvements across software engineering, web development and knowledge-intensive tasks. Google describes it as its most intelligent "workhorse" model for coding and agents to date.
A major focus of Gemini 3.7 Flash is software development. Google said the model delivers better first-pass code accuracy and improved debugging and issue resolution compared with its predecessor.
On FrontierCode 1.1 Main, a benchmark measuring production code quality, Gemini 3.7 Flash scored 43.6%, compared with 34.4% for Gemini 3.6 Flash. On DeepSWE v1.1, another software engineering evaluation, the new model scored 65.3% against 49% for the previous version. These figures are benchmark results cited by Google and should not be interpreted as universal measures of real-world performance.
Google has also targeted web development. The company said Gemini 3.7 Flash can generate more functional layouts and feature-complete applications with fewer prompts, while improving its adherence to screenshots, images and design systems supplied as references.
The model recorded an Elo score of 1,588 on Arena.ai's WebDev Arena, up from 1,538 for Gemini 3.6 Flash.
Beyond coding, Google is positioning the model for knowledge-heavy business applications. It reported improvements in processing complex documents across areas including finance, law and biosciences.
On AutomationBench, which evaluates real-world business workflow completion, Gemini 3.7 Flash scored 30.4%, compared with 17% for Gemini 3.6 Flash.
The model has also been designed to perform multi-step planning and tool calls with greater consistency. Google said it can adapt better when encountering roadblocks, follow instructions more closely and reduce the number of retries and manual interventions required during engineering workflows.
Pricing is another part of Google's positioning. Through the end of 2026, Gemini 3.7 Flash will cost $0.75 per million input tokens and $3.75 per million output tokens. Google said the introductory rate is half the original per-token cost of Gemini 3.6 Flash.
Gemini 3.7 Flash is being made available to developers through the Gemini API, Google AI Studio, Android Studio and Google Antigravity. Enterprise customers can access it through the Gemini Enterprise Agent Platform and Gemini Enterprise app.
Google is also bringing the model to Gemini Spark, its personal AI agent available to Google AI Pro and Ultra subscribers in more than 160 countries. Spark can carry out multi-step tasks across Google Workspace applications, including consolidating files, drafting emails and updating documents.
The company said Gemini 3.7 Flash includes updated safeguards covering cyber misuse and chemical, biological, radiological and nuclear risks.
The launch reflects the growing emphasis among AI developers on models designed not only to answer prompts, but also to execute longer workflows using tools. With Gemini 3.7 Flash, Google is pairing that agentic push with lower introductory API costs as developers and enterprises increasingly evaluate AI on both capability and deployment economics.