NEW DELHI: Google is giving its AI workhorse one other improve, and this one comes with a less expensive price ticket too.
Google has launched Gemini 3.7 Flash, which it describes as its most clever Flash mannequin but for coding and AI brokers. The discharge comes simply three weeks after Gemini 3.6 Flash and is aimed toward enhancing software program engineering, data work and net improvement workflows.
In line with Google, Gemini 3.7 Flash delivers stronger efficiency on advanced coding duties resembling debugging and situation decision, whereas additionally enhancing first-pass code accuracy and production-ready code technology.
On the FrontierCode 1.1 Essential benchmark, the mannequin scored 43.6 per cent, in contrast with 34.4 per cent for Gemini 3.6 Flash. On DeepSWE v1.1, it scored 65.3 per cent in opposition to 49 per cent for its predecessor.
The mannequin additionally claims beneficial properties in net improvement. Google mentioned Gemini 3.7 Flash can create extra useful layouts and feature-complete functions with fewer prompts, whereas higher matching reference screenshots, photos and design methods.
On Enviornment.ai’s WebDev Enviornment, the brand new mannequin recorded an Elo rating of 1,588, in contrast with 1,538 for Gemini 3.6 Flash.
Google can also be positioning Gemini 3.7 Flash for knowledge-intensive fields resembling finance, legislation and biosciences.
On the GDP.pdf benchmark, which assessments a mannequin’s skill to course of advanced paperwork, Gemini 3.7 Flash scored 34 per cent, in contrast with 22 per cent for Gemini 3.6 Flash.
It additionally posted a 30.4 per cent rating on AutomationBench, in opposition to 17 per cent for the sooner mannequin, indicating stronger efficiency on real-world enterprise workflows.
The corporate mentioned the mannequin is best at adapting when it encounters roadblocks, clarifying intent and following directions. It additionally places larger effort into multi-step planning and power calls, probably lowering the necessity for guide intervention and repeated makes an attempt.
The mannequin’s different headline characteristic is its pricing.
Gemini 3.7 Flash will probably be accessible via the tip of 2026 at an introductory value of $0.75 per million enter tokens and $3.75 per million output tokens.
Google mentioned the mix of improved efficiency and decrease value may permit builders and companies to scale production-ready AI brokers extra economically.
Google can also be bringing Gemini 3.7 Flash to Gemini Spark, its private AI agent accessible to Google AI Professional and Extremely subscribers in additional than 160 nations.
Spark, launched at Google I/O, is designed to function repeatedly below consumer course and take actions on their behalf. With the brand new mannequin, Google mentioned Spark will turn into extra environment friendly at data work and power use throughout Google Workspace.
The improve will permit Spark to deal with duties resembling consolidating recordsdata, drafting emails and updating standing paperwork with improved accuracy and output high quality.
Google mentioned Gemini 3.7 Flash is delivery with up to date safeguards protecting potential misuse involving chemical, organic, radiological and nuclear threats, in addition to cyber offences.
The corporate mentioned the up to date protections are supposed to help helpful makes use of whereas strengthening its defences in opposition to dangerous functions.
With Gemini 3.7 Flash, Google is betting that the subsequent leap in AI won’t merely come from making fashions smarter. It can come from making them extra able to finishing advanced work, requiring fewer prompts and costing much less to run.
