Gemini 3.7 Flash: Entry Price at $0.75 and Updated Spark

Google launches Gemini 3.7 Flash with introductory rates of $0.75 per 1M input tokens and $3.75 per 1M output tokens, accessible to developers, businesses, and individuals through its platforms. The model powers Gemini Spark for AI Pro and Ultra subscribers starting today, while updated CBRN and cyber safeguards are announced.

Google sets $0.75 per 1M input and deploys access across its platforms

Gemini 3.7 Flash is offered at $0.75 per 1M input tokens and $3.75 per 1M output tokens until the end of the year. Businesses can access it through the Gemini Enterprise Agent Platform and the Gemini Enterprise app, while individuals can access it via Spark in the Gemini app for Google AI Pro and Ultra subscribers in supported countries. For developers, Google states that it is possible to explore agent workflows in Google Antigravity and get started today via the Gemini API in Google AI Studio and Android Studio, with a dedicated guide. According to Google, this pricing level, combined with performance, should enable the scalable production of agents in a cost-effective manner, and initial customer feedback indicates better results than 3.6 Flash at a reduced cost.

3.7 Flash outperforms 3.6 Flash on FrontierCode, DeepSWE, WebDev Arena, GDP.pdf, and AutomationBench

On FrontierCode 1.1 Main, Google reports 43.6% for 3.7 Flash compared to 34.4% for 3.6 Flash. On DeepSWE v1.1, 3.7 Flash achieves 65.3% against 49.0%. In web development, the Elo score displayed on Arena.ai’s WebDev Arena is 1588 for 3.7 Flash compared to 1538 for 3.6 Flash. For document-heavy tasks, the GDP.pdf benchmark indicates 34.0% against 22.0%. In AutomationBench, Google cites 30.4% compared to 17.0% for completing business workflows.

Google announces advancements in coding, web, and expert domains

Google reports advancements in coding, particularly in debugging and problem-solving, with improved accuracy on the first pass. In web development, 3.7 Flash is said to produce complete applications and more efficient layouts with fewer requests, while faithfully adhering to design and ensuring equivalence from reference elements like screenshots, images, or complete design systems. In fields where knowledge is particularly demanded, such as finance, law, or biosciences, the publisher notes improvements in reasoning and accuracy. Google also highlights an enhanced developer experience: better adaptation to obstacles, clarification of intents, increased instruction tracking, as well as more advanced planning and better-orchestrated tool calls, aimed at reducing manual oversight and retries.

Gemini Spark switches to 3.7 Flash for AI Pro and Ultra subscribers

Gemini Spark now utilizes 3.7 Flash and remains available to Google AI Pro and Ultra subscribers in over 160 countries. Launched at the I/O event as a personal agent operating 24/7 under user direction, Spark is presented as more efficient for knowledge work and tool usage in Google Workspace, with improved accuracy and output quality for complex, multi-skill workflows. With 3.7 Flash, Google states that Spark can consolidate files, draft emails, and update status documents more effectively.

Google claims the most powerful Flash model and updates security

Google presents 3.7 Flash as its smartest working model to date for coding and agents, released three weeks after 3.6 Flash. The publisher attributes this iteration to developer feedback and algorithmic innovations, highlighting substantial improvements in software engineering, knowledge work, and web development, with an introductory price announced as half the original cost of 3.6 Flash per million tokens. On the safety front, Google says it is continuously enhancing the coverage and robustness of its Frontier Safety measures; 3.7 Flash comes with updated protections against CBRN abuse and cyberattacks, which Google presents as aligned with its bio-resilience approach and cyber program.