Google DeepMind announced the release of Gemini 3.7 Flash, its latest AI model designed for coding and agentic workflows, on August 13, 2026. This update arrives an unusually short three weeks after the Gemini 3.6 Flash release, reflecting a rapid development cycle driven by developer feedback and algorithmic advancements. The company is positioning 3.7 Flash as its "most intelligent workhorse model yet," emphasizing its capabilities in software engineering, knowledge work, and web development.
Gemini 3.7 Flash introduces substantial improvements in coding tasks, including debugging and issue resolution. Benchmarks indicate a notable increase in first-pass code accuracy and the generation of production-ready code. On the FrontierCode 1.1 Main benchmark, which assesses production code quality, Gemini 3.7 Flash achieved a score of 43.6%, up from 34.4% for Gemini 3.6 Flash. Similarly, in the DeepSWE v1.1 evaluation for long-horizon software engineering, the new model reached 65.3%, compared to its predecessor's 49.0%. These figures also place Gemini 3.7 Flash ahead of Claude Sonnet 5 (42.7%) and GPT-5.6 Terra (41.3%) on FrontierCode 1.1 Main in Google's comparisons.
For web development, Gemini 3.7 Flash demonstrates improved ability to generate functional layouts and feature-complete applications with fewer prompts. The model also shows higher design adherence and parity when generating user interfaces from various inputs, such as screenshots or design systems. It outperformed 3.6 Flash on Arena.ai's WebDev Arena, achieving an Elo score of 1588 against 1538.
Beyond coding, the model shows gains in knowledge-intensive fields like finance, law, and biosciences, exhibiting improved reasoning and accuracy. On the GDP.pdf benchmark, designed to test a model's ability to process complex documents, 3.7 Flash scored 34.0%, a rise from 22.0% for 3.6 Flash. It also surpassed 3.6 Flash in AutomationBench, an evaluation for real-world business workflow completion, with a score of 30.4% versus 17.0%.
Google DeepMind noted that Gemini 3.7 Flash offers a more refined developer experience. The model is designed to adapt better to obstacles, clarify intent when necessary, and follow instructions with greater precision. It engages in more diligent multi-step planning and tool calls, which the company states can lead to reduced manual oversight and fewer retries in engineering workflows.
To encourage adoption, Google is offering Gemini 3.7 Flash at an introductory price until the end of 2026. The cost is $0.75 per million input tokens and $3.75 per million output tokens, which is half the original price of 3.6 Flash per million tokens. After December 31, 2026, the standard pricing will increase to $1.50 per million input tokens and $7.50 per million output tokens.
Gemini 3.7 Flash maintains a 1 million token context window and supports multimodal inputs including text, images, audio, and video files. It also features customizable thinking configurations, allowing users to adjust the balance between quality, cost, and latency. The model is accessible to developers through the Gemini API in Google AI Studio and Android Studio, as well as Google's Antigravity environment. Enterprise users can access it via the Gemini Enterprise Agent Platform and the Gemini Enterprise app. Additionally, individuals with Google AI Pro or Ultra subscriptions can utilize 3.7 Flash through Spark, the personal agent in the Gemini app.
The release also includes updated safety safeguards to address potential misuse in areas such as Chemical, Biological, Radiological, and Nuclear (CBRN) risks and cyber offense, while enabling beneficial applications. These updates align with Google DeepMind's approach to bioresilience and its cyber program. The rapid release schedule for the Flash series highlights Google's focus on iterative improvements, even as the anticipated Gemini 3.5 Pro model has yet to be publicly released.