Google: Gemini 3.7 Flash Rolled Out in Spark, API, and Professional Tools

The new Gemini 3.7 Flash model is activated in consumer and professional products, including the Spark agent, accessible to subscribers in over 160 countries but absent from several European markets. Google highlights gains in software development and use cases in business, with a measured positioning on a dedicated web benchmark. An introductory price applies until December 31, 2026, before a planned doubling on January 1, 2027.

Spark Welcomes 3.7 Flash in Over 160 Countries, Excluding EEA and the UK

On August 13, Gemini 3.7 Flash was activated in Gemini Spark for Google AI Pro and AI Ultra subscribers in over 160 countries. This availability does not include the European Economic Area, Switzerland, the United Kingdom, or Nigeria. Spark, presented as a personal agent operating continuously from the cloud and functioning within Workspace applications even when the computer is off, was unveiled at Google I/O in May. According to Google, this evolution of the model allows Spark to enhance its efficiency for knowledge tasks while optimizing the use of tools offered in Google Workspace. The company also mentions improved accuracy and quality in complex processes involving multiple skills, such as file management, email drafting, or updating tracking documents.

API, Antigravity, and Professional Tools: Access Points Multiply

Beyond Spark, Google indicates that Gemini 3.7 Flash is already present in several of its products. The model is available in Google Antigravity, in the Gemini API via Google AI Studio, as well as in Android Studio. It is also integrated into the Gemini Enterprise Agent Platform and the Gemini Enterprise application.

Code and Web: Claimed Gains and Ranking on WebDev Arena

Google presents Gemini 3.7 Flash as an advancement in development tasks, from code generation to debugging and ticket resolution. The company claims that in web development, the model produces more functional layouts and more complete applications with fewer prompts, and that it better adheres to designs and reference inputs, whether they are images, screenshots, or design systems. According to Google, the model also manages complex documents better in knowledge-dense fields such as finance, law, and sciences. It is said to be more adept at automating certain professional tasks and planning sequences of actions. On the external measurement side, 3.7 Flash surpasses 3.6 Flash on WebDev Arena by Arena.ai, with an Elo of 1,588 compared to 1,538. However, it is ranked eighth and remains behind Claude Opus 5, Kimi K3, and Grok 4.6. This release follows 3.6 Flash, just three weeks after the latter and 3.5 Flash-Lite.

Transitional Pricing Until December 31, 2026

Google applies an introductory price for Gemini 3.7 Flash until December 31, 2026. During this period, the rate is $0.75 for every million tokens in input and $3.75 for every million tokens in output. From January 1, 2027, the rates are expected to double to match those applied at the launch of Gemini 3.6 Flash.