🔶 Grok 4.6

今天，我们发布 Grok 4.6。Grok 4.6 在 Grok 4.5 的基础上进一步发展，特别聚焦于长时间运行的 Agent，以及更具雄心的交互式和视觉工作。无论是研究某个主题、分析信息、跨代码库工作，还是将一个想法转化为成熟的应用程序或工作成果，它都能在由多个步骤构成的复杂任务中持续推进。

Grok 4.6 在多项 Agent 编程和知识工作基准中实现了前沿智能水平。在 Artificial Analysis Intelligence Index 上，它与 GPT-5.6 Sol 持平；该指数是由九项基准组成的综合评分。

图表所涉基准： AA Intelligence、GDPVal-AA、DeepSWE 1.1、CursorBench 3.2、FrontierCode 1.1。

text 0 20 40 60 AA Intelligence Index 62 Fable 5 Max 61 Grok 4.6 61 GPT-5.6 Sol Max 56 Grok 4.5 High

竞品数据取自相应开发者公开发布的系统卡或基准排行榜。

对比 Grok 4.6 与其他领先模型在 AA Intelligence、GDPVal-AA、DeepSWE 1.1、CursorBench 3.2 和 FrontierCode 1.1 上表现的基准柱状图。竞品数据取自相应开发者公开发布的系统卡或基准排行榜。

Grok 4.6 现已在 Cursor 和 Grok Build 中上线。为让你立即开始试用 4.6，我们将在第一周为 Grok Build 和 Cursor 提供 2 倍包含用量。

与 Grok 4.5 相比，Grok 4.6 进行了更长时间的补充训练，使用了为推理和高级技术概念筛选的模型生成数据、高质量工程数据，以及改进后的优化器和训练配方。这为后续的 SFT 和 RL 阶段奠定了更强的基础。

随后，我们使用 Grok 4.5，针对推理工作、Agent harness 以及 STEM、软件工程和知识工作等领域，重新生成了 SFT 轨迹；并通过基于模型的检查过滤掉有问题的轨迹。由此得到的 SFT 检查点展现出强劲性能和更好的行为表现。

Grok 4.6 在广泛的 Agent RL 任务上接受训练，包括知识工作、通用编程，以及面向内核优化、Web 开发、计算机辅助设计等领域的专用环境。

我们在旨在拓展其能力范围、并测试其能否在多个步骤中持续工作的项目上测试了 Grok 4.6。我们发现，该模型尤其擅长将宽泛的产品构想转化为可运行的首个版本。它可以研究陌生领域、构建应用结构、实现核心交互，并在多轮反馈中持续完善结果。

在更长的任务轨迹中，我们还开始看到更多自我测试和验证行为：模型会在继续下一步之前检查自己的工作。

与我们通常在 Grok 4.5 上观察到的情况相比，Grok 4.6 能为视觉和交互项目产出更强的首版结果。面对具体的产品想法，它能够在一次生成中确立应用的结构和视觉语言。这使其特别适合这类项目：最快获得良好结果的路径，是先从一个足够完整的成果开始，再在循环中不断迭代。

Grok 4.6 的安全防护措施已得到改进，并根据模型能力进行了校准。

我们的安全体系旨在合法使用场景中最大化实用性与安全性，使 Grok 4.6 能够在漏洞修补、加速工程设计周期、增强 AI 研究等领域既有帮助又足够安全。

我们的安全防护评估工作反映了 Grok 4.6 扩展后的能力：我们开展了迄今覆盖范围最广的部署前能力测试和安全防护校准测试，以及广泛的部署后和第三方测试。

模型 / 评测项	Grok 4.5 High	GPT-5.6 Sol Max	Fable 5 Max
AA Intelligence Index	56	61
CursorBench v3.2	66.7%	67.2%
FrontierCode v1.1（Extended）	56.6%	60.6%
Terminal-Bench v3.0	15.7%	34.1%
Harvey LAB（Vals）	12.9%	2.5%	11.3%

每项评测中的最佳分数以粗体标示。第三方模型分数采用自行报告结果或公开可获得结果中的最佳值。

Grok 4.6 现已在 Cursor 和 Grok Build 中上线。它也已通过 API 提供，并已接入 OpenRouter、Vercel 和 Cloudflare 等合作伙伴。

定价为每百万输入 token 2 美元、每百万输出 token 6 美元。此外，还有一个快速版本，价格为上述价格的两倍。

为让你立即开始试用 4.6，我们将在第一周为 Grok Build 和 Cursor 提供 2 倍包含用量。

创建 API 密钥

立即通过 SpaceXAI API 开始使用 Grok 4.6 进行开发。

API 文档

阅读文档，并将 Grok 4.6 集成到你的技术栈中。

在 Grok Build 中免费试用

立即前往 x.ai/build 开始使用。

bash $ curl -fsSL https://x.ai/cli/install.sh | bash

Today we are releasing Grok 4.6. Grok 4.6 builds on Grok 4.5 with a particular focus on long-running agents and more ambitious interactive and visual work. It stays with complex tasks across many steps, whether researching a topic, analyzing information, working across a codebase, or turning an idea into a polished application or work artifact.

Grok 4.6 achieves frontier intelligence across several agentic coding and knowledge work benchmarks. It matches GPT-5.6 Sol on the Artificial Analysis Intelligence Index, which is a composite score of nine benchmarks.

AA IntelligenceGDPVal-AADeepSWE 1.1CursorBench 3.2FrontierCode 1.1

0204060AA Intelligence Index62Fable 5 Max61Grok 4.661GPT-5.6 Sol Max56Grok 4.5 High

Competitor figures are drawn from the respective developers’ published system cards or benchmark leaderboards

Benchmark bar charts comparing Grok 4.6 with other leading models across AA Intelligence, GDPVal-AA, DeepSWE 1.1, CursorBench 3.2, and FrontierCode 1.1. Competitor figures are drawn from the respective developers’ published system cards or benchmark leaderboards.

Grok 4.6 is available today in Cursor and Grok Build. We’re offering 2x included usage inside Grok Build and Cursor for the first week so you can start trying 4.6 immediately.

Grok 4.6 underwent a longer supplemental training run than Grok 4.5, with curated model-generated data for reasoning and advanced technical concepts, high-quality engineering data, and an improved optimizer and training recipe. This produced a stronger foundation for the SFT and RL stages that followed.

We then used Grok 4.5 to regenerate the SFT trajectories across reasoning efforts, agent harnesses, and domains such as STEM, software engineering, and knowledge work, and filtered out problematic traces with model-based checks. The resulting SFT checkpoint shows strong performance and improved behavior.

Grok 4.6 is trained on a wide range of agentic RL tasks, including knowledge work, general coding, and domain-specific environments for kernel optimization, web development, computer-aided design, and more.

We tested Grok 4.6 on projects designed to stretch its range and ability to sustain work over many steps. We found the model is especially strong at turning a broad product idea into a working first version. It can research unfamiliar domains, structure the application, implement the core interactions, and continue refining the result through several rounds of feedback.

On longer trajectories, we also started to see more self-testing and verification, with the model checking its own work before moving on.

Grok 4.6 produces stronger first passes on visual and interactive projects than we typically saw with Grok 4.5. Given a concrete product idea, it is able to establish structure and visual language for an application in one pass. This has made it especially useful for projects where the fastest route to a good result was to begin with something substantial and then iterate in the loop.

Grok 4.6’s safeguards have been improved and calibrated in line with the model’s capabilities.

Our safety stack is designed to maximize utility and security across legitimate use cases, allowing Grok 4.6 to be helpful and safe in domains such as vulnerability patching, accelerating the engineering design cycle, and augmenting AI research.

Our safeguard evaluation work reflects Grok 4.6’s expanded capabilities, with our widest-ever suite of pre-deployment testing for capabilities and safeguard calibration, as well as extensive post-deployment and third-party testing.

Grok 4.5 High

GPT-5.6 Sol Max

Fable 5 Max

AA Intelligence Index

56

61

CursorBench v3.2

66.7%

67.2%

FrontierCode v1.1 (Extended)

56.6%

60.6%

Terminal-Bench v3.0

15.7%

34.1%

Harvey LAB (Vals)

12.9%

2.5%

11.3%

Best score per evaluation in bold. Third-party model scores are the best of self-reported or publicly available results.

Grok 4.6 is available today in Cursor and Grok Build. It’s also available in the API and other partners like OpenRouter, Vercel, and Cloudflare.

Pricing starts at $2 per million input tokens and $6 per million output tokens. Additionally, there is a fast variant which is twice the price.

We’re offering 2x included usage inside Grok Build and Cursor for the first week so you can start trying 4.6 immediately.

Create an API Key

Start building with Grok 4.6 today via the SpaceXAI API.

API Docs

Read the docs and integrate Grok 4.6 into your stack.

Try it in Grok Build for free

Get started today at x.ai/build.

$ curl -fsSL https://x.ai/cli/install.sh | bash