2026年7月21日
|
15 分钟阅读
我们最新的 Gemini 模型提供了构建大规模 AI 智能体所需的效率、延迟和可靠性。
收听文章
[[duration]] 分钟
此内容由 Google AI 生成。生成式 AI 为实验性技术。
构建生产级 AI 智能体的开发者和客户需要更高的 Token 效率、更低的延迟和更可靠的性能。我们的 Flash 系列模型旨在满足效率与质量的平衡点，从而支持智能体工作流的规模化扩展。在 Gemini 3.5 Flash 的基础上，我们推出了新的 Gemini 模型：
- 3.6 Flash： 我们的主力模型，在编码、知识工作和多模态性能方面表现更佳。根据 Artificial Analysis Index，与 3.5 Flash 相比，其输出 Token 使用量减少了 17%，并且在某些基准测试（如 Datacurve 的 DeepSWE）中，我们观察到减少幅度高达 65%，同时每个输出 Token 的成本更低。
- 3.5 Flash-Lite： 我们最快、最具成本效益的 3.5 级模型，根据 Artificial Analysis Index，每秒可输出 350 个 Token，在智能体工作流中的表现也显著优于之前的 Flash-Lite 代际。
- CodeMender 中的 3.5 Flash Cyber： 成功的网络安全应用需要将模型与智能体基础设施进行精心编排。我们推出了一款新型、高效、专注于网络安全的专用模型，并将其与我们的 CodeMender 代码安全智能体相结合，在前沿领域提供了具有竞争力的性能。
除了今天的发布之外，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尽快广泛提供。与此同时，我们的团队已经在专注于构建下一代模型。我们已经开始了迄今为止最雄心勃勃的预训练运行，即 Gemini 4，并对取得的进展感到兴奋。
3.6 Flash：比 3.5 Flash 更高效、质量更优
Gemini 3.6 Flash 直接基于开发者和客户对 3.5 Flash 的反馈构建。3.6 Flash 不仅在编码和知识工作方面实现了提升，同时还显著提高了 Token 效率。例如，在 Artificial Analysis Index 上，我们看到 3.6 Flash 消耗的输出 Token 比 3.5 Flash 少 17%。它完成多步骤工作流所需的推理步骤和工具调用也更少。
这种增强的效率还与低于 3.5 Flash 的价格相结合。3.6 Flash 的输入 Token 价格为每百万个 1.50 美元，输出 Token 价格为每百万个 7.50 美元，降低了每个智能体任务的总成本，使智能体的构建和运行更具成本效益。
在 OSWorld 验证任务（API）中，3.6 Flash 显示出比 3.5 Flash 更好的 Token 效率和更低的冗长程度
即使在更高效的同时，3.6 Flash 在多个用例中的性能相比 3.5 Flash 也有所提升：
- 3.6 Flash 提供了更高的精度，减少了不必要的代码编辑和执行循环，如 DeepSWE（49% 对比 37%）所示，并在 ML 研究方面表现出显著改进，如 MLE Bench（63.9% 对比 49.7%）所示。
- 其计算机使用能力得到提升，如 OSWorld-Verified（83.0% 对比 78.4%）所示。计算机使用现在通过 Gemini API 和 Gemini Enterprise 成为内置的客户端工具。
- 在知识工作方面，它优于 3.5 Flash，如 GDPval-AA v2（1421 对比 1349）等基准测试所示。Hebbia 和 Harvey 等客户发现它在文档解析、图表和数据分析以及报告起草等多模态任务中特别有能力。
客户报告称，3.6 Flash 在成本和质量方面都向前迈进了一步，在复杂工作流和基于知识的任务中平衡了 Token 效率、准确性和速度：
安全构建
3.6 Flash 在化学、生物、放射性和核（CBRN）以及网络攻击滥用领域配备了增强的前沿安全保障措施。这些保障措施使模型更能抵御越狱攻击。同时，该模型经过训练，可最大限度地减少对有益用途的拒绝。
有关更多信息，请参阅 3.6 Flash 模型卡。
3.5 Flash-Lite：为扩展智能体工作流而构建
除了 Flash，我们还发布了 Gemini 3.5 Flash-Lite，专为低延迟任务以及高吞吐量对开发者工作流至关重要的任务（如智能体搜索和文档处理）而设计。
3.5 Flash-Lite 是 3.5 系列中最快的模型。根据 Artificial Analysis 的测量，其运行速度为每秒 350 个输出 Token。定价为输入 Token 每百万个 0.3 美元，输出 Token 每百万个 2.5 美元，并且质量显著优于 3.1 Flash-Lite，3.5 Flash-Lite 为运行高吞吐量生产流量的开发者和客户提供了强大的性价比。
3.5 Flash-Lite 以比 3.5 Flash 更低的延迟执行高容量任务。
3.5 Flash-Lite 实现了智能体系统的高效扩展。在各个思考级别上，该模型显著优于 3.1 Flash-Lite。根据工作负载的不同，开发者可以配置模型，使用最低和低思考级别优先处理高容量任务的低延迟、低成本执行，或使用更高的思考级别来处理多步骤子智能体工作负载。该模型现在还将计算机使用作为内置工具，以可靠地支持跨表面的这些智能体任务。
在编码和智能体任务方面，如 Terminal-Bench 2.1（54% 对比 31%）所示；在长上下文方面，如 GDM-MRCR v2（72.2% 对比 60.1%）所示；以及在现实世界任务执行方面，如 GDPval-AA v2（1140 对比 642）所示，它都是一个显著的进步。
事实上，在许多智能体和编码评估中，3.5 Flash-Lite 甚至优于 3 Flash，包括在 SWE-Bench Pro（54.2% 对比 49.6%）和 OSWorld-Verified（74.0% 对比 65.1%）上，使其成为 2.5 和 3 Flash 工作负载更快、更有能力的选择。
3.5 Flash-Lite 的早期客户强调了其在扩展智能体工作流和数据处理任务方面独特的速度、智能和成本效益组合：
有关该模型的更多信息，请参阅 3.5 Flash-Lite 模型卡。
CodeMender 中的 3.5 Flash Cyber：高效发现和修复漏洞
AI 模型已经能够比当前系统修复漏洞的速度更快地发现安全漏洞。应对这一日益增长的威胁需要一种高度有能力和高效的软件安全方法。
Flash 的性能与效率使其成为大规模检测、验证及修补代码安全问题的理想基础。Gemini 3.5 Flash Cyber 基于 3.5 Flash 构建，并针对发现和修复网络安全漏洞进行了微调，其每 token 价格低于更大模型。
在 CodeMender（该工具通过多个 3.5 Flash Cyber 智能体协同工作，生成单一综合报告）中，3.5 Flash Cyber 在热门基准测试 CyberGym 上达到了前沿水平的竞争力。
鉴于该技术的双重用途特性，我们采取了审慎的方式部署 3.5 Flash Cyber。该模型将很快通过 CodeMender 独家提供给政府和受信任合作伙伴，作为有限访问试点计划的一部分。这将使一线防御者能够在关键漏洞被利用之前抢先发现并修复它们，同时降低更广泛的滥用风险。
3.6 Flash 与 3.5 Flash-Lite：立即开始使用
3.6 Flash 和 3.5 Flash-Lite 即日起可用：
- 面向开发者：通过 Google AI Studio 和 Android Studio 在 Gemini API 中使用。3.6 Flash 也可在 Google Antigravity 中使用。请参阅开发者指南开始使用。
- 面向企业：在 Gemini Enterprise Agent Platform 中使用。3.6 Flash 也可在 Gemini Enterprise 应用中使用。
- 面向所有用户：通过 Gemini 应用使用。3.5 Flash-Lite 也正在 Google 搜索中逐步推出。
当您开始使用 3.6 Flash 和 3.5 Flash-Lite 构建时，我们欢迎您提供反馈以改进未来的 Gemini 模型，并期待很快发布 3.5 Pro。
Jul 21, 2026
|
15 min read
Our newest Gemini models deliver the efficiency, latency, and reliability to build AI agents at scale.
Listen to article
[[duration]] minutes
This content is generated by Google AI. Generative AI is experimental
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:
- 3.6 Flash: Our workhorse model that delivers better coding, knowledge work, and multimodal performance. According to the Artificial Analysis Index, it reduces output token usage by 17% compared to 3.5 Flash, and in some benchmarks like DeepSWE by Datacurve, we observe up to 65%, all at a lower cost per output token.
- 3.5 Flash-Lite: Our fastest, most cost-effective 3.5-class model, delivering 350 output tokens per second according to the Artificial Analysis Index, also significantly outperforming prior Flash-Lite generations in agentic workflows.
- 3.5 Flash Cyber in CodeMender: Successful cybersecurity applications require careful orchestration of a model alongside an agent infrastructure. We’re introducing a combination of a new, highly efficient, specialized cyber-focused model paired with our CodeMender code security agent that delivers competitive performance at the frontier.
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.
3.6 Flash: More efficient and better quality than 3.5 Flash
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 delivers higher precision with fewer unwanted code edits and reduced execution loops, as seen in DeepSWE (49% vs. 37%), and shows significant improvement in ML Research, as seen in MLE Bench (63.9% vs. 49.7%).
- It has improved computer use capabilities as seen in OSWorld-Verified (83.0% vs. 78.4%). Computer use is now a built-in client side tool via the Gemini API and Gemini Enterprise.
- It outperforms 3.5 Flash in knowledge work, as shown by benchmarks like GDPval-AA v2 (1421 vs. 1349). Customers like Hebbia and Harvey have found it particularly capable at multimodal tasks like document parsing, chart and data analysis, and report drafting.
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:
Built with safety
3.6 Flash is shipping with enhanced Frontier Safety 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 model card.
3.5 Flash-Lite: Built to scale agentic workflows
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, 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.
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 model card.
3.5 Flash Cyber in CodeMender: finding and fixing vulnerabilities efficiently
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 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 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: Get started today
3.6 Flash and 3.5 Flash-Lite are available starting today:
- For developers in the Gemini API via Google AI Studio and Android Studio. 3.6 Flash is also available in Google Antigravity. Get started with the Developer Guide.
- For enterprises in Gemini Enterprise Agent Platform. 3.6 Flash is also available in the Gemini Enterprise app.
- For everyone via the Gemini app. 3.5 Flash-Lite is also rolling out in Google Search.
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.
本文内容采集自官方网站，排版和翻译可能与原页面存在差异。
阅读官方全文