Google introduces Gemini 3.7 Flash for coding and agents

Google introduced Gemini 3.7 Flash on August 13, calling it its most intelligent workhorse model yet for coding and agents. The release arrives only three weeks after Gemini 3.6 Flash and is available now through the Gemini API in Google AI Studio, Android Studio, Google Antigravity and Google’s enterprise products. Individuals can access it through Gemini Spark on eligible Google AI Pro and Ultra subscriptions.

Google reports substantial gains in software engineering and web development. On its published tests, Gemini 3.7 Flash scored 43.6% on FrontierCode 1.1 Main, compared with 34.4% for 3.6 Flash, and 65.3% versus 49.0% on DeepSWE v1.1. Its WebDev Arena Elo score rose from 1538 to 1588. Google also reports improvements in complex document work and business automation. These are company-presented benchmarks, so independent testing remains important.

The commercial change is unusually concrete. Through the end of 2026, Google is offering an introductory price of $0.75 per million input tokens and $3.75 per million output tokens. That is half the original price of Gemini 3.6 Flash. Google has not stated the price that will apply after the introductory period, which matters for teams estimating the long-term cost of production agents.

Google says the model is better at adapting when a workflow hits a roadblock, asking for clarification when needed and following instructions across multi-step plans and tool calls. Gemini Spark, the company’s always-running personal agent, is switching to 3.7 Flash in more than 160 countries. Google says this should improve work with Workspace tools, including consolidating files, drafting emails and updating status documents.

The release also includes updated safeguards against misuse in chemical, biological, radiological and nuclear domains and in offensive cybersecurity. Google provides further details in a model card, while access routes differ for developers, enterprises and individual subscribers.

For AI makers, the rapid release cadence shows how quickly model economics and agent performance are moving together. Better benchmark scores and lower introductory pricing could make longer coding and knowledge-work workflows more practical, but reliability, latency, security and the post-2026 price still need real-world evaluation. For users, the most visible effect will come through Spark. For businesses, Gemini 3.7 Flash is a reason to retest existing agent workloads rather than assume that a newer model automatically produces a safer or cheaper deployment.