Google DeepMind 发布 Gemini 3.6 Flash、3.5 Flash-Lite 与 3.5 Flash Cyber 三款新模型

开发人员和客户在构建生产级 AI 代理时，需要更高的令牌效率、更低的延迟以及更可靠的性能。我们的 Flash 系列模型旨在实现效率与质量的最佳平衡，从而支持代理工作流程的扩展。在 Gemini 3.5 Flash 的基础上，我们推出了新的 Gemini 模型：

除了当前发布的版本外，Gemini 3.5 Pro 目前正在与合作伙伴进行测试，我们计划在准备就绪后广泛提供。同时，我们的团队已经开始专注于构建下一代模型。我们已启动了迄今为止最雄心勃勃的预训练运行，用于 Gemini 4，并对进展感到非常兴奋。

Gemini 3.6 Flash 直接基于开发人员和客户对 3.5 Flash 的反馈。3.6 Flash 不仅在编码和知识工作方面有所提升，同时显著提高了令牌效率。例如，在人工分析指数（Artificial Analysis Index）上，我们发现 3.6 Flash 的输出令牌比 3.5 Flash 少 17%。在完成多步骤工作流程时，它还需要更少的推理步骤和工具调用。

这种增强的效率还伴随着比 3.5 Flash 更低的价格。3.6 Flash 的价格为每 100 万输入令牌 $1.50，每 100 万输出令牌 $7.50，从而降低了每项代理任务的总体成本，使构建和运行代理更加具有成本效益。

在 OSWorld 验证任务（API）中，3.6 Flash 显示出比 3.5 Flash 更高的令牌效率和更少的冗长。

即使效率更高，3.6 Flash 在各用例中相比 3.5 Flash 也显示出性能提升：

3.6 Flash 使用 AIS 上的托管代理，可以比 3.5 Flash 更高效、更准确地解析和分析财务数据及会议记录（AIS）

3.6 Flash 在 AGY 上使用多代理编排执行代码迁移，具有比 3.5 Flash 更低的延迟和更高的质量（AGY）

3.6 Flash 使用 canvas（Gemini 应用）帮助开发 3D 工作流程的摄影纹理提取器

3.6 Flash 利用 AGY 和 tldraw 离线编辑器，凭借其强大的视觉理解能力构建交互式主题工作室（AGY）

客户表示 3.6 Flash 在成本和质量上都迈出了一步，在复杂工作流程和基于知识的任务中平衡了令牌效率、准确性和速度：

3.6 Flash 搭载了增强的前沿安全措施：https://deepmind.google/blog/strengthening-our-frontier-safety-framework/，涵盖化学、生物、放射性和核（CBRN）及网络攻击误用领域的防护。这些防护措施使模型在防止越狱方面大大增强。同时，模型也经过训练，以最大限度减少对有益用途的拒绝。

更多信息请参见 3.6 Flash 的模型卡：https://deepmind.google/models/model-cards/gemini-3-6-flash/。

除了 Flash，我们还发布了 Gemini 3.5 Flash-Lite，专为低延迟任务以及对开发者工作流中高吞吐率关键的任务而设计，如智能搜索和文档处理。

3.5 Flash-Lite 是 3.5 系列中最快的模型。根据 Artificial Analysis 的测量：https://artificialanalysis.ai/models/gemini-3-5-flash-lite，它的处理速度为 350 个输出标记/秒。价格为 $0.3/100 万输入标记和 $2.5/100 万输出标记，并且质量明显优于 3.1 Flash-Lite。3.5 Flash-Lite 为处理高吞吐量生产流量的开发者和客户提供了强有力的性价比。

3.5 Flash-Lite 在高负载任务的执行延迟上低于 3.5 Flash。

3.5 Flash-Lite 可实现智能系统的高效扩展。在各级思考能力上，该模型显著优于 3.1 Flash-Lite。根据工作负载，开发者可以配置模型优先处理具有最低和低思考级别的高吞吐量任务，以实现低延迟、低成本执行；或者启用更高的思考级别来处理多步骤子代理工作负载。模型现在还内置计算机工具，以可靠地支持这些智能任务。

在编码和智能任务方面取得了显著提升，如 Terminal-Bench 2.1（54% 对 31%）、长上下文能力见于 GDM-MRCR v2（72.2% 对 60.1%），以及实际任务执行见于 GDPval-AA v2（1140 对 642）。

实际上，在许多智能和编码评测中，3.5 Flash-Lite 甚至优于 3 Flash，包括 SWE-Bench Pro（54.2% 对 49.6%）和 OSWorld-Verified（74.0% 对 65.1%），使其在 2.5 和 3 Flash 的工作负载下成为更快速且更强大的选择。

3.5 Flash-Lite 从庞大的电子商务数据集中提取产品特征并进行综合。

在 3.6 Flash 作为主代理的配合下，3.5 Flash-Lite 可以瞬间生成 25 个独特的、可立即探索的网页设计概念。

3.5 Flash-Lite 可以利用其多模态理解能力扩展收据翻译和摘要功能。

3.5 Flash-Lite 通过即时生成并迭代多种选项来构建游戏。

3.5 Flash-Lite 的早期客户强调了其在扩大代理工作流和数据处理任务中的速度、智能与成本效率的独特结合：

有关该模型的更多信息，请参见 3.5 Flash-Lite：https://deepmind.google/models/model-cards/gemini-3-5-flash-lite/ 模型卡。

人工智能模型已经能够比当前系统修复速度更快地发现安全漏洞。应对这一不断增长的威胁需要一种高度能力和高效的软件安全方法。

Flash 的性能和效率使其成为大规模检测、验证和修补代码安全问题的理想基础。Gemini 3.5 Flash Cyber：https://deepmind.google/blog/introducing-gemini-3-5-cyber-flash/ 构建在 3.5 Flash 之上，并经过微调，以较低的每令牌成本发现和修复网络安全漏洞，相较于更大的模型更经济。

在 CodeMender 中，多个 3.5 Flash Cyber 代理协作生成单一合并报告，3.5 Flash Cyber 在流行基准测试 CyberGym 上达到了前沿的竞争性能。

鉴于该技术的双重用途性质，我们在部署 3.5 Flash Cyber 时采取了谨慎的方法。该模型将通过 CodeMender：https://deepmind.google/blog/introducing-codemender-an-ai-agent-for-code-security/ 限量访问试点程序，仅向政府和可信合作伙伴提供。这将使前线防御者在漏洞被利用前获得先机来发现和修复关键漏洞，同时减少更广泛的滥用风险。

3.6 Flash 和 3.5 Flash-Lite 自今日起可用：

在您开始使用 3.6 Flash 和 3.5 Flash-Lite 构建时，我们欢迎您的反馈以改进未来的 Gemini 模型，并期待尽快发布 3.5 Pro。

检查你的收件箱以确认你的订阅。

让我们保持联系。通过你的收件箱获取谷歌的最新消息。

Developers and customers building production AI agents need higher token efficiency, lower latency, and more reliable performance. Our Flash series of models is built to meet the sweet spot of efficiency and quality to enable scaling agentic workflows. Building on Gemini 3.5 Flash, we’re introducing new Gemini models:

Beyond today’s releases, Gemini 3.5 Pro is currently testing with partners and we plan to make it broadly available as soon as it’s ready. In parallel, our team is already focusing on building the next generation of models. We have started our most ambitious pre-training run yet, for Gemini 4, and are excited by the progress.

Gemini 3.6 Flash builds directly on developer and customer feedback from 3.5 Flash. 3.6 Flash not only delivers a step up in coding and knowledge work, but it does this while meaningfully improving token efficiency. For example, on the Artificial Analysis Index, we see 3.6 Flash consuming 17% fewer output tokens than 3.5 Flash. It also takes fewer reasoning steps and tool calls to accomplish multi-step workflows.

This enhanced efficiency is also combined with a lower price than 3.5 Flash. At $1.50/1M input tokens and $7.50/1M output tokens, 3.6 Flash reduces the overall cost per agentic task, making agents more cost-effective to build and run.

3.6 Flash shows better token efficiency and reduced verbosity than 3.5 Flash in an OSWorld verified task (API)

Even while being more efficient, 3.6 Flash sees performance gains compared to 3.5 Flash across use cases:

3.6 Flash, using Managed Agents on AIS, can help parse through and analyze financial data and transcripts more efficiently and accurately than 3.5 Flash (AIS)

3.6 Flash executes code migrations, using multi-agent orchestration on AGY, with lower latency and higher quality than 3.5 Flash (AGY)

3.6 Flash helps develop a photographic texture extractor for 3D workflows, using canvas (Gemini App)

3.6 Flash uses AGY and the tldraw offline editor to build interactive theme studios with its strong visual understanding skills (AGY).

Customers report 3.6 Flash is a step forward in both cost and quality, balancing token efficiency, accuracy, and speed across complex workflows and knowledge-based tasks:

3.6 Flash is shipping with enhanced Frontier Safety：https://deepmind.google/blog/strengthening-our-frontier-safety-framework/ safeguards in the domains of Chemical, Biological, Radiological, and Nuclear (CBRN) and cyber offense misuses. These safeguards make the model substantially more resistant to jailbreaks. At the same time, the model has been trained to minimize refusals for beneficial uses.

For more information, see the 3.6 Flash：https://deepmind.google/models/model-cards/gemini-3-6-flash/ model card.

Beyond Flash, we’re also releasing Gemini 3.5 Flash-Lite, designed for both low-latency tasks and tasks where high throughput is critical for developers workflows, like agentic search and document processing.

3.5 Flash-Lite is the fastest model in the 3.5 series. As measured by Artificial Analysis：https://artificialanalysis.ai/models/gemini-3-5-flash-lite, it runs at 350 output tokens/s. Priced at $0.3/1M input tokens and $2.5/1M output tokens and with significantly better quality than 3.1 Flash-Lite, 3.5 Flash-Lite offers a strong price-to-performance ratio for developers and customers running high throughput production traffic.

3.5 Flash-Lite executes high volume tasks at a lower latency than 3.5 Flash.

3.5 Flash-Lite enables efficient scaling for agentic systems. Across thinking levels, the model significantly outperforms 3.1 Flash-Lite. Depending on the workload, developers can configure the model to prioritize low-latency, low-cost execution for high-volume tasks with the minimal and low thinking levels, or engage higher thinking levels to process multi-step subagent workloads. The model now also has computer use as a built-in tool to reliably support these agentic tasks across surfaces.

It’s a significant step up in coding and agentic tasks as seen in Terminal-Bench 2.1 (54% vs 31%), long context as seen in GDM-MRCR v2 (72.2% vs. 60.1%), and real-world task execution as seen in GDPval-AA v2 (1140 vs. 642).

In fact, on many agentic and coding evals, 3.5 Flash-Lite even outperforms 3 Flash, including on SWE-Bench Pro (54.2% vs. 49.6%) and OSWorld-Verified (74.0% vs. 65.1%), making it a faster & more capable option for workloads on both 2.5 and 3 Flash.

3.5 Flash-Lite extracts product features from a massive e-commerce dataset and synthesizes it.

Working alongside 3.6 Flash as the master agent, 3.5 Flash-Lite instantly generates 25 unique, ready-to-explore web design concepts.

3.5 Flash-Lite can scale receipt translation and summarization with its multimodal understanding.

3.5 Flash-Lite builds a game by instantly generating and iterating through multiple options.

Early customers of 3.5 Flash-Lite are highlighting its unique combination of speed, intelligence, and cost efficiency for scaling agentic workflows and data processing tasks:

For more information about the model, see the 3.5 Flash-Lite：https://deepmind.google/models/model-cards/gemini-3-5-flash-lite/ model card.

AI models have become capable of finding security vulnerabilities faster than current systems can fix them. Tackling this growing threat requires an approach to securing software that is highly capable and efficient.

Flash’s performance and efficiency makes it an ideal foundation to detect, validate, and patch code security issues at scale. Gemini 3.5 Flash Cyber：https://deepmind.google/blog/introducing-gemini-3-5-cyber-flash/ is built on top of 3.5 Flash, and fine-tuned for finding and fixing cybersecurity vulnerabilities at a lower price per token than larger models.

Within CodeMender, which uses multiple 3.5 Flash Cyber agents working together to produce a single combined report, 3.5 Flash Cyber reaches competitive performance at the frontier on the popular benchmark CyberGym.

Given the dual-use nature of this technology, we have taken an intentional approach to deploying 3.5 Flash Cyber. The model will be exclusively available to governments and trusted partners via CodeMender：https://deepmind.google/blog/introducing-codemender-an-ai-agent-for-code-security/ soon as part of a limited-access pilot program. This will give frontline defenders a head start in finding and fixing critical vulnerabilities before they can be exploited, while mitigating against broader misuse.

3.6 Flash and 3.5 Flash-Lite are available starting today:

As you start building with 3.6 Flash and 3.5 Flash-Lite, we welcome your feedback to improve future Gemini models and look forward to releasing 3.5 Pro soon.

Check your inbox to confirm your subscription.

Let's stay in touch. Get the latest news from Google in your inbox.