--- 格式版本: 2 标题: "Introducing Gemini 3.8 Flash and 3.8 Flash Cyber" 原文链接: "https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber/" 发布日期: "2026-09-02" 发布时间校准状态: "found" 发布时间需复核: "否" 发布时间来源: "rule:scrape:strict_html_metadata" 发布时间证据: "article:published_time: 2026-09-02" 发布时间校准原因: "规则确认唯一严格发布时间，来源 scrape:strict_html_metadata" 发布时间校准置信度: "high" 发布时间候选数量: 10 发布时间严格候选数量: 3 发布时间原页读取状态: "source template page reused from URL open" 发布时间未找到原因: "" 发布时间校准时间: "2026-09-07T22:17:51+08:00" 发布时间仲裁状态: "skipped" 发布时间仲裁尝试次数: 0 发布时间仲裁耗时毫秒: 0 发现时间: "2026-09-07T22:07:26+08:00" 入库时间: "2026-09-07T14:39:40.356Z" 来源平台: "固定入口" 搜索渠道: "fixed_url" 搜索词: "https://blog.google/" 匹配关键词: - "AI" - "deployment" - "performance" 相关厂家: - "Google" 相关专家: [] 内容类型: "网页" 抓取工具: "Free Fetch + Defuddle" 清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取" 原始附件: [] AI优质: "否" AI打分: 53 AI分档: "非优质" AI质检状态: "不通过" AI打分理由: "正文主线仅为Gemini 3.8 Flash及3.8 Flash Cyber大模型发布、评测与可用渠道，未讨论超节点、AI Rack、机架级系统或其关键部件；来源为Google官方博客，权威但与机架级基础设施主题无关；历史对照：与知识库中Google TPU、AI Hypercomputer、NVLink/MGX等机架级事实无直接重复，本文不含机架架构、互连、供电、液冷或RAS增量；新增事实为新模型版本、Cyber变体、定价与评测数据；未命中高价值准入通道，命中应用与模型效率否决项。" AI质检模型: "zj-deepseek-v4-flash" AI质检时间: "2026-09-07T22:40:37+08:00" AI主题相关性: 2 AI来源权威性: 14 AI新颖性: 16 AI技术细节: 5 AI商业部署信号: 8 AI完整性: 8 AI评分提示词版本: "v17-精简生产版" AI评分提示词SHA256: "48fb9777f386026761b4873eaff30807694fb11e9b352d7c69bf2dfde750cc7d" AI评分知识库版本: "knowledge_base_v1-20260819+runtime.99" AI评分知识库SHA256: "0fbd91b8efdf08ea0580fc2df2684b6db7e286b65c7b6430d7666940f4e954c3" AI评分知识库检索词: "[\"Google\",\"https://blog.google/\",\"RAS\",\"NPU\",\"Intel\",\"footnote-1\",\"width-2000.format-webp_66O8lls.webp\",\"v1.1\",\"width-2000.format-webp.webp\",\"V2\",\"www.vals.ai/benchmarks/fabv2\",\"HLE-Verified\"]" AI评分知识库命中: "[{\"id\":\"historical-jan-apr-02\",\"title\":\"二、Google Cloud Next '26：AI Hypercomputer 与第八代 TPU 发布\",\"sourceType\":\"curated_item\",\"time\":\"2026-01_to_2026-04\",\"matchedTerms\":[\"Google\",\"Intel\"],\"rank\":-9.950281178091092},{\"id\":\"july-correct-0112\",\"title\":\"超前点映AMD Advancing AI 2026：AMD AI YES？-36氪\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"Intel\",\"V2\"],\"rank\":-8.686816197261713},{\"id\":\"july-correct-0001\",\"title\":\"全球首颗2nm GPU来了！苏姿丰甩出“最强AI机架”，CPU性能干翻英伟达 - 智东西\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"RAS\",\"NPU\"],\"rank\":-7.55028615485027},{\"id\":\"runtime-613d1ef67a8028d2dc2916c4\",\"title\":\"Nvidia, MediaTek Bring Custom Chips to AI Racks\",\"sourceType\":\"ai_excellent_article\",\"time\":\"2026-08-31\",\"matchedTerms\":[\"Google\",\"RAS\",\"Intel\"],\"rank\":-6.7121378005934496},{\"id\":\"runtime-f15ede7e8a5c3e80bd078fbd\",\"title\":\"Ultra-High Interactivity on NVIDIA GPUs? - TileRT InferenceX\",\"sourceType\":\"ai_excellent_article\",\"time\":\"2026-08-10\",\"matchedTerms\":[\"Google\",\"RAS\",\"NPU\",\"V2\"],\"rank\":-6.675693569150028}]" 采集批次: "2026年9月7日22点05分18秒" 采集批次ID: "20260907-220517-361" 去重键: "https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber" --- Building on the momentum of [3.7 Flash](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-3-7-flash/) from three weeks ago and marking our third Flash release in only six weeks, today we’re introducing Gemini 3.8, our best reasoning and coding model yet, at the same speed and low cost of 3.7. Gemini 3.8 introduces 2 variants: - **Gemini 3.8 Flash:** our most intelligent workhorse model, delivering significant improvements from 3.7 Flash across software engineering, agentic tasks, and critical, multi-step reasoning in specialized domains. It is available at the same introductory price [1](#footnote-1) as 3.7 Flash at $0.75 per million input tokens and $3.75 per million output tokens. - **Gemini 3.8 Flash Cyber:** our most capable cybersecurity model with frontier-level performance in vulnerability detection and automated patching, available to trusted defenders through our new [Fairwind Program](https://deepmind.google/fairwind-program/). While tailored for different deployment environments, both of today's releases are powered by the same foundational intelligence, and further accelerated by long-running agentic loops designed to recursively evaluate and refine the underlying models. The significant coding and reasoning gains across this shared core were driven by a number of innovations, including rigorous training in the highly demanding domain of cybersecurity. ## Gemini 3.8 Flash: built for long-horizon coding and autonomous agents Gemini 3.8 Flash delivers substantial gains from 3.7 Flash, often approaching the performance of higher-cost frontier models. ![comparison chart showing Gemini 3.8 Flash and other models](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3.8-flash__evals__table_l.width-2000.format-webp_66O8lls.webp) **On DeepSWE v1.1 (Long-Horizon Software Engineering**) 3.8 Flash outperforms most larger frontier models in autonomously solving complex engineering problems end to end, only at a fraction of the cost. ![chart showing Gemini 3.8 Flash DeepSWE v1.1](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3-8-flash__evals__deepswe.width-2000.format-webp.webp) Additionally, 3.8 Flash exhibits the dependability required for critical enterprise autonomy, across specialized knowledge domains**.** In quantitative and professional fields that require advanced analysis and reporting, 3.8 Flash outperforms 3.7 Flash and other frontier models in benchmarks like [Vals Finance Agent V2](https://www.vals.ai/benchmarks/fabv2) and [Harvey's Legal Agent Benchmark](https://www.vals.ai/benchmarks/hlab). 3.8 Flash also achieves a 54.9% on HLE-Verified, demonstrating its ability to handle multi-step reasoning across STEM, humanities, and professional fields. ![chart showing Gemini 3.8 Flash Vals Finance Agent v2](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3.8-flash_evals_vals-finan.width-100.format-webp.webp) ![a chart showing Gemini 3.8 Flash Harvey's Legal Agent Benchmark](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3.8-flash_evals_harveys-le.width-100.format-webp.webp) ![a chart showing Gemini 3.8 Flash HLE-Verified](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3.8-flash_evals_HLE_light.width-100.format-webp.webp) These performance gains stem from a core design choice: 3.8 Flash works harder. On complex tasks, it exhibits greater diligence — executing extra reasoning steps, and calling tools iteratively. At times, the model might use more tokens to maximize performance, especially at higher effort levels. For applications where compute efficiency is the primary constraint, developers can utilize lower effort levels to minimize token overhead or continue to rely on Gemini 3.7 Flash, which remains fully supported for efficiency-first workloads. Gemini 3.8 Flash built this game with a simple prompt using a looping instruction in Google Antigravity. The game uses puzzles, environmental storytelling, and textures generated with Nano Banana to create an immersive 3D level in which you play a wizard navigating a castle. Gemini 3.8 Flash builds a fully functional DOS version of Google Maps in a single prompt in Google Antigravity that is fully playable with locations, directions, and Street View. Explore realtime cross-sections, 2D projections, scientific explanations in a topographic map of famous geographical sites built with Gemini 3.8 Flash in Google Antigravity using real datasets from the U.S. Geological Survey. Hardware Anatomy is an interactive 3D visualizer built with Gemini 3.8 Flash in Google AI Studio that generates realistic Three.js renderings of physically-proportioned teardowns for hardware devices. It automatically decomposes devices into layers you can explode and inspect with a deconstruction slider. ## Gemini 3.8 Flash Cyber: expert cyber performance Gemini 3.8 Flash Cyber, available to a set of trusted defenders via the [Fairwind Program](https://blog.google/innovation-and-ai/technology/safety-security/fairwind-program), provides a decisive advantage in today’s complex cybersecurity landscape, with the Flash speed and cost that enables quick iteration. ### Autonomous vulnerability discovery On the standard industry benchmark for finding vulnerabilities, CyberGym, Gemini 3.8 Flash Cyber demonstrates frontier-level performance in autonomous vulnerability discovery. It surpasses both 3.5 Flash Cyber as well as significantly larger frontier models. ![a chart showing Gemini 3.8 Flash Cyber CyberGym Pass@1](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3.8-cyber__evals__cybergy.width-2000.format-webp.webp) To better capture real-world defensive needs which are not limited to just C/C++ codebases like in CyberGym, we also evaluated Gemini 3.8 Flash Cyber against a comprehensive internal benchmark in which the model has to discover a wide range of vulnerabilities across complex codebases spanning 20 programming languages. Here, the model showcases an impressive leap over our previous models and reaches a success rate exceeding 70%. ![Chart showing Gemini 3.8 Flash Cyber real world vulnerability discovery](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3.8-cyber__evals__rw-vuln.width-2000.format-webp.webp) ### Automated patching With Gemini 3.8 Flash Cyber, we focused specifically on equipping defenders with expert capabilities that give them an advantage over attackers. This is why we have invested in vulnerability fixing from the start, and prioritized it over offensive capabilities like exploitation. [CWE-Bench](https://cwe-bench.com/#leaderboard), run by Collinear, is a challenging external benchmark for patching capabilities. On this benchmark, Gemini 3.8 Flash Cyber is on the Pareto frontier: with a pass@1 of 47.2% compared to a leading frontier model at 47.8%, yet offered at a significantly lower cost. ![Chart showing Gemini 3.8 Flash Cyber StaticBench Pass@1 vs Cost Per Rollout](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3-8-cyber__evals__cwe-ben.width-2000.format-webp.webp) ### Real-world impact: securing Google’s code We’re already using Gemini 3.8 Flash Cyber to secure code across Google. For example: - The Chrome Security team found that 3.8 Flash Cyber produced 2.6 times more correct patches to vulnerabilities in Chrome than the best commercial models that are much larger. - Wiz found that Gemini 3.8 Flash Cyber achieves +7.5-9.7% higher recall on their internal penetration testing benchmark for a 2.3-5.2x lower cost compared to other leading frontier models. - Google’s Cloud Vulnerability Research team leveraged the 3.8 Flash Cyber model to find a critical foundational vulnerability in less than 2 hours, a vulnerability for which research and discovery usually takes months. ## What our Fairwind Program partners are saying ![quote from David Slater, Founder and Chief Architect, Armadin](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3-8-flash__testimonial__ar.width-100.format-webp.webp) ![quote from Charlie Sestito, Director, Office of the CTO, Palo Alto Networks](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3-8-flash__testimonial__pa.width-100.format-webp.webp) ![quote from Mayank Upadhyay, CSTO, Snowflake](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3-8-flash__testimonial__sn.width-100.format-webp.webp) ![quote from Gal Nagli, Head of Threat Exposure, Wiz](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3-8-flash__testimonial__wi.width-100.format-webp.webp) ## Built with safety in mind 3.8 Flash ships with safeguards against misuse in the domains of Chemical, Biological, Radiological, and Nuclear (CBRN) and cyber offense, while enabling beneficial use cases, as per our [Frontier Safety Framework](https://deepmind.google/blog/strengthening-our-frontier-safety-framework/). 3.8 Flash Cyber ships with a more permissive set of mitigations for cybersecurity, and as such, is only available to trusted defenders who require a more comprehensive set of cyber capabilities. Gemini 3.8 models have also made a significant leap in prompt injection robustness as measured by Gray Swan, protecting Gemini model users from prompt-injection related malicious attacks. ![a chart showing Gemini 3.8 Flash Cyber Gray Swan IPI Benchmark](https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3.8-flash_evals_attack__l.width-2000.format-webp.webp) ## Gemini 3.8 Flash and Cyber: get started today - **Developers**: Build with 3.8 Flash and explore agent-first workflows in [Google Antigravity](https://antigravity.google/) or start building today in the Gemini API via [Google AI Studio](https://aistudio.google.com/prompts/new_chat?model=gemini-3.8-flash) and [Android Studio](https://developer.android.com/studio), or generate UIs in [Stitch](http://stitch.withgoogle.com/). Get started with our [developer docs](https://ai.google.dev/gemini-api/docs/latest-model). - **Enterprises**: Access 3.8 Flash in [Gemini Enterprise.](https://console.cloud.google.com/agent-platform/studio/multimodal?mode=prompt&model=gemini-3.8-flash) - **Consumers**: 3.8 Flash is available to Google AI Pro and Ultra subscribers across the [Gemini app](http://gemini.google.com/), [AI Mode](http://google.com/ai) in Google Search and Gemini in [Google Sheets](http://sheets.new/). - **Cyber:** Through our new [Fairwind Program](https://blog.google/innovation-and-ai/technology/safety-security/fairwind-program), we’re providing trusted government authorities, as well as critical infrastructure operators and software maintainers with prioritized access to Gemini 3.8 Flash Cyber. [Apply for access](https://deepmind.google/fairwind-program/).