Google 發布全新 Gemini 3.7 Flash AI 模型
Google Releases New Gemini 3.7 Flash AI Model
Updated at: August 14, 2026 at 01:15 AM
2026年8月13日，Google正式發布了Gemini 3.7 Flash，這是其人工智慧模型陣容的最新成員。
On August 13, 2026, Google officially launched Gemini 3.7 Flash, the latest addition to its AI model lineup.
此模型被定位為多功能主力，專門為處理軟體工程、網頁開發和代理工作流程中的複雜任務而設計。
Positioned as a versatile workhorse, this model is specifically engineered to handle complex tasks in software engineering, web development, and agentic workflows.
基於Gemini 3.6 Flash的成功，新版本在性能上有顯著提升，特別是在DeepSWE v1.1等程式編寫基準測試以及各類推理任務中。
Building on the success of Gemini 3.6 Flash, the new version delivers significant performance boosts, particularly in coding benchmarks like DeepSWE v1.1 and various reasoning tasks.
Google將開發重點放在改進模型的推理基礎上，使其能以更高的準確性和效率進行多步驟規劃和工具使用。
Google focused its development on refining the model's reasoning foundation, allowing it to navigate multi-step planning and tool usage with greater accuracy and efficiency.
儘管有這些進步，Google仍保持了該模型的成本效益，並提供特別的推廣定價至2026年底。
Despite these advancements, Google has kept the model cost-effective, offering special introductory pricing until the end of 2026.
目前該模型已在整個Google生態系統中開放，包括AI Studio、Gemini API和GitHub Copilot。
The model is now available across the entire Google ecosystem, including AI Studio, the Gemini API, and GitHub Copilot.
Gemini 3.7 Flash擁有超過一百萬個代幣（Token）的龐大上下文視窗，並支援文本、圖像和音訊的多模態功能，標誌著Google快速更新週期中向前邁出了重要一步，確保開發者能擁有強大、安全且可靠的工具，以建構下一代人工智慧驅動的應用程式。
With a massive context window of over one million tokens and robust multimodal support for text, images, and audio, Gemini 3.7 Flash represents a significant step forward in Google's rapid update cycle, ensuring developers have access to a powerful, safe, and reliable tool for building the next generation of AI-driven applications.
