Google发布Gemini 4 Neon模型
Gemini 4 Neon
全新旗舰模型发布,性能与成本优势显著,开发者应重点关注其工程落地能力与定价策略。
Introducing Gemini 4 Neon
介绍 Gemini 4 Neon
Nov 18, 2026
2026年11月18日
- x.com
- Copy link
- x.com
- 邮件
- 复制链接
Bringing frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows.
在复杂的软件工程、知识工作和科学工作流中,实现快速端到端任务完成的前沿级智能。
Koray Kavukcuoglu
SVP, Google DeepMind and Chief AI Architect, Google
Google DeepMind 高级副总裁兼 Google 首席 AI 架构师
Share
分享
- x.com
- Copy link
- x.com
- 邮件
- 复制链接
Your browser does not support the audio element.
您的浏览器不支持音频元素。
Listen to article
收听文章
[[duration]] minutes
[[duration]] 分钟
This content is generated by Google AI. Generative AI is experimental
此内容由 Google AI 生成。生成式 AI 处于实验阶段
Voice Speed
语音速度
Voice
语音
Speed 0.75X 1X 1.5X 2X
速度 0.75X 1X 1.5X 2X
In this article
在本文中
Today, we’re introducing Gemini 4 Neon — bringing frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows.
今天,我们推出 Gemini 4 Neon——在复杂的软件工程、知识工作和科学工作流中,以前沿级智能实现快速端到端任务完成。
Neon ranks ahead of comparable models and #3 overall with a score of 61 on the Artificial Analysis Intelligence Index at its xhigh thinking level, landing just behind larger, slower models. On the Vals Index, Neon ranks #2 and lands on the Pareto frontier. Beyond just pushing on raw model capability, we’re continuing to make our models better to work with, Neon ranks #2 on steerability (adapting cleanly to mid-task user corrections) on the new Agent Arena Index. Neon also excels at multi-step instruction following and builds on our ongoing progress to curb fabrication, recording our lowest fabrication rate yet in internal Google Antigravity testing.
在 Artificial Analysis Intelligence Index 的 xhigh 思考级别下,Neon 以 61 分的成绩优于同类模型,总体排名第 3,仅次于更大、更慢的模型。在 Vals Index 上,Neon 排名第 2,并位于帕累托前沿(Pareto frontier)。除了提升原始模型能力外,我们还在持续优化模型的易用性;在新发布的 Agent Arena Index 上,Neon 在可操控性(steerability,即适应任务中途用户修正的能力)方面排名第 2。Neon 在多步指令遵循方面也表现出色,并在抑制虚构内容方面延续了我们的持续进展,在 Google Antigravity 内部测试中创下了最低的虚构率记录。
The model is priced at $3 per million input tokens ($0.15 per million cached input tokens, a 95% discount) and $15 per million output tokens, with a 1 million-token context window.
该模型定价为每百万输入令牌 3 美元(缓存输入令牌每百万 0.15 美元,折扣 95%),输出令牌每百万 15 美元,上下文窗口为 100 万令牌。
Gemini 4 Neon is rolling out today in the Gemini API in Google AI Studio, Gemini for business (via API and model dropdown), for paid Google AI Pro and Ultra subscribers in Google Antigravity, and for Pro and Ultra subscribers in the Gemini app.
Gemini 4 Neon 今日开始在 Google AI Studio 的 Gemini API、Google for Business(通过 API 和模型下拉菜单)、Google Antigravity 中的付费 Google AI Pro 和 Ultra 订阅者,以及 Gemini 应用中的 Pro 和 Ultra 订阅者中推出。
Delivers across enterprise, analytical, and scientific work
在企业、分析和科学工作中全面交付
Enterprise work lives in spreadsheets, PDFs, charts, audio, and video. Neon can synthesize across all of these inputs in a single 1 million-token context window and turns inputs into structured deliverables. Independent evaluations rank Neon as one of the top models and on benchmarks like the Vals Index, it ranks #2 (70.9%), and reaches #1 on both the Legal Research Bench and Tax Agent Bench.
企业工作涉及电子表格、PDF、图表、音频和视频。Neon 可以在单个 100 万令牌的上下文窗口中对所有这些输入进行综合处理,并将输入转化为结构化交付物。独立评估将 Neon 列为顶级模型之一;在 Vals Index 等基准测试中,它排名第 2(70.9%),并在 Legal Research Bench 和 Tax Agent Bench 上均达到第 1 名。
Neon also shows strong wins across categories on Surge AI, ranking #1 on most benchmarks across finance, accounting, spreadsheet modeling, and complex constraint following. It also scores 73.5% on Chartography for reading dense professional charts and performs on par with leading models on Agents' Last Exam v1.1 (75.8%). It achieves #1 overall on both Zapier’s AutomationBench (56.3%) and Andon Labs' Vending-Bench 2 ($16,843), a long-horizon simulation of running a business. Finally, it debuts #2 globally with a score of 1336 on Arena's new Agent Arena Index across 4,200+ real-world agent sessions.
Neon 在 Surge AI 的各个类别中也表现出强劲优势,在金融、会计、电子表格建模和复杂约束遵循方面的多数基准测试中均排名第 1。它在 Chartography 上得分 73.5%,用于阅读密集的专业图表;在 Agents' Last Exam v1.1 上的表现与领先模型持平(75.8%)。它在 Zapier 的 AutomationBench(56.3%)和 Andon Labs 的 Vending-Bench 2(16,843 美元,一项模拟经营企业的长周期测试)上均获得全球第 1 名。最后,在包含 4,200 多个真实世界代理会话的新 Agent Arena Index 上,它以 1336 分的成绩在全球排名第 2。
On scientific and mathematical reasoning, spanning quantitative analysis, mathematics, and technical research problems, Neon is best in class on Terminal-Bench Science (70.5%), RiemannBench (84.0%), and HLE-Diamond (71.3%).
在涵盖定量分析、数学和技术研究问题的科学与数学推理方面,Neon 在 Terminal-Bench Science(70.5%)、RiemannBench(84.0%)和 HLE-Diamond(71.3%)上均处于同类最佳水平。
Built for production-ready code
专为生产就绪代码构建
Engineered for long-running agent workflows, Neon sustains performance through complex, tool-heavy tasks end-to-end — whether running standalone or acting as a fast, high-judgment Orchestrator directing parallel Hydrogen subagents in Google Antigravity. It cuts down on redundant file reads, shares progress as it works, and handles context compaction reliably in long sessions.
Neon 专为长期运行的智能体工作流而设计,能够在端到端的复杂、重工具任务中保持性能——无论是独立运行,还是作为快速且具备高判断力的编排器(Orchestrator),指挥 Google Antigravity 中的并行 Hydrogen 子智能体。它能减少冗余的文件读取,在工作过程中共享进度,并在长会话中可靠地处理上下文压缩。
Across core developer and researcher workflows, Neon achieves frontier performance and is competitive with the leading models in its class, ranking #1 overall on the Artificial Analysis Coding Agent Index (58.6), with a score of 65.2% on Terminal-Bench 4.1 and 46.2% (at its default high setting) on FrontierCode 2.0. For production-ready code, Neon handles multi-file debugging and refactoring across repositories with targeted, surgical edits. On SWE-sweep, it already leads on the mini-SWE harness (12.8%) and more than triples to a state-of-the-art 39.0% in Google Antigravity showing what the Antigravity harness unlocks.
在核心开发者和研究人员的工作流程中,Neon 实现了前沿性能,并与同类领先模型具有竞争力,在 Artificial Analysis Coding Agent Index 中总体排名第一(58.6),在 Terminal-Bench 4.1 上得分为 65.2%,在 FrontierCode 2.0 上得分为 46.2%(默认高设置)。针对生产就绪代码,Neon 能够跨仓库进行多文件调试和重构,通过精准的外科手术式编辑完成任务。在 SWE-sweep 基准测试中,它已在 mini-SWE 测试套件中领先(12.8%),并在 Google Antigravity 中将成绩提升至创纪录的 39.0%,展示了 Antigravity 测试套件的潜力。
Driving Blender MCP in Google Antigravity, Gemini 4 Neon used /swarm to build a photorealistic 3D lab bench visualizing the model names with elemental accuracy pairing Bunsen flames for Gemini 4 Carbon and Gemini 4 Hydrogen with high-voltage gas discharge tubes for Gemini 4 Argon and Neon.
在 Google Antigravity 中驱动 Blender MCP 时,Gemini 4 Neon 使用 /swarm 构建了一个照片级真实的 3D 实验室工作台,以元素级的精度可视化模型名称:为 Gemini 4 Carbon 和 Gemini 4 Hydrogen 搭配本生灯火焰,为 Gemini 4 Argon 和 Neon 搭配高压气体放电管。
In Google Antigravity, Gemini 4 Neon wrote a first-person WebGL2 freeride ski simulator from scratch on real Mont Blanc and Chamonix elevation terrain — combining 10,200 lines of custom shaders, 120 Hz ski physics across six alpine lines, dynamic carved tracks, slab avalanches, and spatial Web Audio with zero pre-made assets.
在 Google Antigravity 中,Gemini 4 Neon 从零开始编写了一个第一人称 WebGL2 自由式滑雪模拟器,基于真实的勃朗峰(Mont Blanc)和霞慕尼(Chamonix)地形数据——结合了 10,200 行自定义着色器、六条高山线路的 120 Hz 滑雪物理引擎、动态刻痕轨迹、板状雪崩以及空间 Web Audio,且未使用任何预制资产。
Using /swarm in Google Antigravity, Gemini 4 Neon pulled archival photos to recreate Alan Turing’s 1936 study at King’s College, Cambridge in interactive 3D — from King’s Chapel outside to Room X.8 inside, complete with a working Universal Turing Machine running on his desk.
在 Google Antigravity 中使用 /swarm,Gemini 4 Neon 调取档案照片,以交互式 3D 形式重现了艾伦·图灵(Alan Turing)1936 年在剑桥国王学院的研究场景——从外部的国王教堂到内部的 X.8 房间,桌上还有一台正在运行的通用图灵机。
In Google Antigravity, a Gemini 4 Neon agent swarm turned a single image into a 100,000-cell Google Sheet with zero embedded images and zero Apps Script — using over 500,000 live spreadsheet formulas to run edge detection, map 3D depth, and relight the scene across six interactive modes.
在 Google Antigravity 中,一个 Gemini 4 Neon 智能体群将单张图片转换为包含 100,000 个单元格的 Google Sheet,其中零嵌入图片和零 Apps Script——利用超过 500,000 个实时电子表格公式,在六种交互模式下执行边缘检测、映射 3D 深度并重新打光场景。
Brings efficiency and control to production-grade workflows
为生产级工作流程带来效率与控制力
We’re refining thinking levels with each new model release to give teams finer control and a smoother progression from one tier to the next. Neon’s four levels scale steadily, letting teams dial intelligence and token spend to what each workload needs.
我们在新模型发布中不断优化思维层级,为团队提供更精细的控制以及从一级到下一级的更平滑过渡。Neon 的四个层级稳步扩展,让团队能够根据每项工作负载的需求调整智能水平和 Token 消耗。
Neon delivers high performance at a lower cost per task across third-party benchmarks. At its default high thinking level, it pairs a high Artificial Analysis Intelligence Index score with a 32% lower cost per task than Carbon. It also sits on the Pareto frontier on both the Vals index and Agent Arena at high, with roughly 50% to 80% lower cost per task than top competing frontier models. xhigh thinking adds deeper, multi-step reasoning for the hardest engineering, analytical, and scientific workloads. Neon shows its highest intelligence score here, while still costing 36%+ less per task than competing models at their highest setting.
在第三方基准测试中,Neon 以更低的任务成本提供高性能。在其默认的高思维层级下,它拥有较高的 Artificial Analysis Intelligence Index 评分,且任务成本比 Carbon 低 32%。在高思维层级下,它在 Vals 指数和 Agent Arena 上均处于帕累托前沿(Pareto frontier),任务成本比顶级竞争的前沿模型低约 50% 至 80%。xhigh 思维层级针对最复杂的工程、分析和科学工作负载提供更深度的多步推理。在此层级,Neon 展现出最高的智能评分,同时其任务成本仍比竞争对手在最高设置下的成本低 36% 以上。
Built for fast, efficient, frontier-level cyber defense
专为快速、高效的前沿级网络防御而构建
As agentic coding scales, generating secure code from the start is critical. Gemini 4 Neon for cyber produces more secure code than Carbon across all easy, medium, and hard challenges on AutoBaxBench, achieving state-of-the-art on the hard tier (76%).
随着代理式编码规模的扩大,从一开始就生成安全代码至关重要。Gemini 4 Neon for cyber 在 AutoBaxBench 的所有简单、中等和困难挑战中生成的安全代码均优于 Carbon,并在困难层级达到最先进的水平(76%)。
Securing systems also means proactively finding flaws before attackers do. Penetration testing has served as a cornerstone of enterprise cybersecurity. On WebExploitBench v1.2, Neon is state-of-the-art (51.26%) at half the cost of competing large models.
保护系统也意味着在攻击者之前主动发现漏洞。渗透测试一直是企业网络安全的基石。在 WebExploitBench v1.2 上,Neon 以竞争大型模型一半的成本达到了最先进的水平(51.26%)。
To fortify existing code, Neon for cyber delivers frontier-level performance across the code hardening lifecycle — vulnerability finding, validating, and patching — all while being faster and more cost efficient than large frontier models.
为了加固现有代码,Neon for cyber 在代码加固生命周期——漏洞发现、验证和修补——中提供前沿级性能,同时速度更快且比大型前沿模型更具成本效益。
- Finding code vulnerabilities: Neon achieves state-of-the-art performance on SEC-bench Pro (90.5%), while costing roughly half as much per challenge.
- Validating risk through exploitation: To prioritize which vulnerabilities to remediate, defenders rely on models to check which vulnerabilities are exploitable. On ExploitGym, it achieves frontier-level performance using half the tokens of Carbon.
- Patching vulnerabilities: Fixing the root cause of vulnerabilities is a critical step in the cybersecurity defense lifecycle. Neon reaches 70.7% on Collinear’s CWE-bench v2, driving a 3.4% uplift over Carbon.
- 发现代码漏洞:Neon 在 SEC-bench Pro 上达到最先进的性能(90.5%),而每个挑战的成本约为其他模型的一半。
- 通过利用来验证风险:为了优先处理哪些漏洞需要修复,防御者依赖模型来检查哪些漏洞可被利用。在 ExploitGym 上,它使用仅为 Carbon 一半的 Token 量即实现了前沿级性能。
- 修补漏洞:修复漏洞的根本原因是网络安全防御生命周期中的关键步骤。Neon 在 Collinear 的 CWE-bench v2 上达到 70.7%,较 Carbon 提升了 3.4%。
Gemini 4 Neon for cyber is rolling out to trusted defenders in the coming days as part of the Fairwind Program.
Gemini 4 Neon for cyber 将在未来几天作为 Fairwind Program 的一部分向受信任的防御者推出。
Designed for safe, dependable execution
专为安全、可靠的执行而设计
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力