Gemini 4 Argon

本文为原文前 6,000 字符的节选翻译，完整内容请查看原文。

Argon helped the Chrome team separate core webpage handling from a ~6 million line legacy folder of browser features held within Chromium’s 40 million-line codebase. Piping multiple clean code changes at a time, Argon helped the team finish an architectural refactor across hundreds of files, completing months of work in weeks. For Chrome’s billions of global users, this will unlock the ability for live webpages to move fluidly between tabs, side panels, and floating windows without reloading, eliminating an entire category of browser crashes.

Argon 帮助 Chrome 团队将核心网页处理功能从 Chromium 4000 万行代码库中一个约 600 万行的遗留浏览器功能文件夹中分离出来。通过同时推送多个整洁的代码变更，Argon 帮助团队完成了跨数百个文件的架构重构，将数月的工作量缩短至数周内完成。对于 Chrome 全球数十亿用户而言，这将使实时网页能够在标签页、侧边栏和浮动窗口之间流畅切换而无需重新加载，从而消除了一整类浏览器崩溃问题。

xTB (Semiempirical Extended Tight-Binding) is one of the most-cited open-source quantum chemistry packages for molecules. Scientists use it to predict molecules’ 3D shapes and to screen large numbers of drug and materials candidates. Using Argon in Antigravity, our researchers made xTB up to 2.3x faster at predicting molecular structures and 1.8x faster at conformer search. The changes have been contributed to the official xTB repository on GitHub.

xTB（半经验扩展紧束缚方法）是分子领域引用率最高的开源量子化学软件包之一。科学家利用它来预测分子的 3D 形状，并筛选大量的药物和材料候选物。通过在 Antigravity 中使用 Argon，我们的研究人员使 xTB 在预测分子结构方面的速度提升了 2.3 倍，在构象搜索方面的速度提升了 1.8 倍。这些更改已贡献给 GitHub 上的官方 xTB 存储库。

Argon in Antigravity is helping the team behind Google’s open-source Agent Development Kit (ADK), a framework powering enterprise agents, push their performance engineering further. The team gave Argon their .agents/skills rulebook and a one-paragraph goal to eliminate framework overhead without breaking existing apps. Working autonomously overnight, Argon surfaced optimizations, now live in google/adk-python, that cut in-memory session prep time by 40% in long sessions and sped up cold imports by 21%.

Antigravity 中的 Argon 正在帮助 Google 开源 Agent Development Kit (ADK) 的团队进一步提升其性能工程水平，该框架为企业级智能体提供支持。团队向 Argon 提供了他们的 .agents/skills 规则手册，并设定了一个简短的目标：在不破坏现有应用程序的前提下消除框架开销。Argon 在夜间自主工作，发现了多项优化方案，现已在 google/adk-python 中上线，这些优化将长会话中的内存会话准备时间缩短了 40%，并将冷导入速度提高了 21%。

Using Argon, we achieved a 2.85x end-to-end inference speedup on an A100 via vLLM (accelerating throughput from ~37,900 to over 108,000 tokens per second on Gemma 4 E2B) with 100% token parity and zero quality drop in just a few days, a process that would usually take months. Check out the upstream pull requests (optimizing everything from attention tuning to speculative decoding) to see the exact changes.

使用 Argon，我们在几天内通过 vLLM 在 A100 上实现了 2.85 倍的端到端推理加速（将 Gemma 4 E2B 的吞吐量从每秒约 37,900 个 token 提升至超过 108,000 个 token），同时保持了 100% 的 token 一致性且质量零下降，而这一过程通常需要数月时间。查看上游的拉取请求（涵盖从注意力机制调整到推测解码的所有优化），即可了解具体的变更内容。

Argon in Antigravity migrated a backlog of 700+ Google Skills training labs off of hardcoded system images and runtimes spread across five code repositories, catching 16+ hidden timeout and race-condition bugs on the way. The work saved 955 engineering hours and reduced recurring cleanup sweeps from roughly 540 hours a year to a single 3-hour update.

Antigravity 中的 Argon 将 700 多个 Google Skills 培训实验室的积压工作从分散在五个代码库中的硬编码系统镜像和运行时中迁移出来，并在此过程中捕获了 16 个以上的隐藏超时和竞态条件错误。这项工作节省了 955 个工程小时，并将每年的重复清理工作从约 540 小时减少到一次 3 小时的更新。

Argon is helping our researchers accelerate progress across both quantum error correction and algorithm design. To help error decoders keep pace with quantum hardware in real-time, Argon translated an internal version of AlphaQubit, our ML quantum error correction decoder, into a single, highly optimized hardware kernel – nearly doubling its inference throughput so we can deploy larger, more capable models and scale to larger quantum devices, while continuing to meet strict real-time thresholds. Argon is also reducing spacetime costs for quantum subroutines. For arcsin, a building block in recent quantum simulation algorithms, an implementation written by Argon and formally verified in Lean cuts spacetime cost by more than 100x compared to a published quantum CORDIC baseline, a step toward a standard library of optimized quantum circuits.

Argon 正在帮助我们的研究人员加速量子纠错和算法设计方面的进展。为了帮助纠错解码器实时跟上量子硬件的步伐，Argon 将我们机器学习量子纠错解码器 AlphaQubit 的内部版本转换为单个高度优化的硬件内核，使其推理吞吐量几乎翻倍，从而使我们能够在满足严格实时阈值的同时，部署更大、能力更强的模型并扩展到更大的量子设备。Argon 还在降低量子子程序的时空成本。对于近期量子模拟算法中的基础模块 arcsin，由 Argon 编写并在 Lean 中经过形式化验证的实现，其时空成本比已发布的量子 CORDIC 基准降低了 100 倍以上，这是迈向优化量子电路标准库的一步。

Stellarators are incredibly complex machines that represent a leading candidate for limitless clean fusion energy, but optimizing their 3D magnetic fields is famously difficult. Since 1967, a long-standing hypothesis known as “Grad’s conjecture” assumed that finding exact mathematical solutions for these fields was impossible, until recent counterexamples were found, including by Professor Matt Landreman (University of Maryland). Now, a team at Google alongside Professor Landreman has discovered a broader class of counterexamples to Grad’s conjecture. These new solutions give the fusion community reliable benchmarks to test their stellarator simulations codes, which are valuable tools for future designs.

仿星器（Stellarators）是极其复杂的机器，代表了实现无限清洁聚变能源的主要候选方案，但优化其 3D 磁场极其困难。自 1967 年以来，一个被称为“Grad 猜想”的长期假设认为，为这些磁场找到精确的数学解是不可能的，直到最近发现了反例，其中包括马里兰大学 Matt Landreman 教授发现的反例。现在，Google 的一个团队与 Landreman 教授合作，发现了 Grad 猜想的一类更广泛的反例。这些新解为聚变界提供了可靠的基准，用于测试其仿星器模拟代码，这些代码是未来设计的宝贵工具。

Researchers at Gamow Labs used Argon to analyze de-identified genomes from newborns with rare developmental lung disorders. In one case, Argon pinpointed a chromosomal rearrangement that routine software missed for years, and explained why it went undetected. On private Gamow Labs evals derived from real clinical and lab data, Argon topped on RareBench (57% on assistance with rare-disease diagnosis) and LabBench (52.2% on wet-lab reasoning). These advances point to faster diagnosis for families facing rare conditions.

Gamow Labs 的研究人员使用 Argon 分析了患有罕见发育性肺部疾病的新生儿的去标识化基因组。在一个案例中，Argon 精确识别出了一种常规软件多年来一直遗漏的染色体重排，并解释了其未被检测到的原因。在基于真实临床和实验室数据的 Gamow Labs 私有评估中，Argon 在 RareBench（罕见病诊断辅助方面达到 57%）和 LabBench（湿实验室推理方面达到 52.2%）中名列前茅。这些进展预示着面临罕见疾病的家庭将能获得更快的诊断。

Engineering enzymes with precise reactivity is notoriously slow, often demanding months of bespoke model tuning and candidate screening. Serine esterases are a class of enzymes that cut ester bonds on targeted molecules, widely used in recycling, and the manufacture of pharmaceuticals, biofuels and detergents. Our scientists worked with Argon to design serine esterases, targeting the benchmark molecule 4MU-acetate that glows under UV light once its ester bonds are broken. In a matter of hours, Argon assembled relevant data, performed model tuning and identified multiple biochemistry-inspired filters to get to a final set of five novel designs. Subsequent laboratory testing confirmed robust catalytic activity across these designs. The speed and ease of successful enzyme generation demonstrates the potential for rapid acceleration in enzyme engineering in fields such as biomanufacturing, green chemistry, and drug development.

设计具有精确反应性的酶过程极其缓慢，通常需要数月的定制模型调整和候选筛选。丝氨酸酯酶是一类能够切割目标分子上酯键的酶，广泛应用于回收利用以及药品、生物燃料和洗涤剂的制造。我们的科学家与 Argon 合作设计了丝氨酸酯酶，目标分子是 4MU-乙酸酯，该分子在酯键断裂后会在紫外线下发光。在短短几个小时内，Argon 汇总了相关数据，执行了模型调整，并确定了多个受生物化学启发的过滤器，最终得出了五种新颖的设计。随后的实验室测试证实，这些设计均表现出强大的催化活性。成功生成酶的速度和便捷性，展示了在生物制造、绿色化学和药物开发等领域加速酶工程的潜力。

Uninterrupted frontier math and science derivations: With 1M output tokens, Argon has the headroom to finish 100K+ token reasoning trajectories and output exhaustive, step-by-step mathematical proofs, quantum error-correction decoders, and complete computational physics solvers in one go.

不间断的前沿数学和科学推导：凭借 100 万个输出 token 的容量，Argon 有足够的空间完成超过 10 万个 token 的推理轨迹，并一次性输出详尽的、分步的数学证明、量子纠错解码器以及完整的计算物理求解器。

Full-repository generation and refactoring: Generate entire multi-file applications, scaffold complete microservices, or refactor sprawling codebases in a single pass without stitching together truncated responses.

全存储库生成与重构：一次性生成整个多文件应用程序、构建完整的微服务框架或重构庞大的代码库，无需拼接截断的响应。

Massive legacy translations: Translate monolithic legacy files and complex database schemas end-to-end while preserving cross-file dependencies.

大规模遗留代码翻译：端到端地翻译单体遗留文件和复杂的数据库架构，同时保留跨文件依赖关系。

Exhaustive structured synthesis: Produce compr

详尽的结构化综合：生成 compr