x.ai发布Grok 4.7：强化编码与长任务能力
Grok 4.7
Grok 4.7是x.ai推出的全新旗舰模型，重点强化了长上下文编码与专业领域知识工作，且给出了极具竞争力的定价与安全指标，值得开发者关注其实际效能。
Back to newsSep 21, 2026
返回新闻2026年9月21日
Introducing Grok 4.7
介绍 Grok 4.7
SpaceXAI's most powerful model for coding and knowledge work. Twice as fast, at half the price of comparable models.
SpaceXAI 最强大的编码与知识工作模型。速度提升两倍，价格仅为同类模型的一半。
Try for freeStart building
免费试用开始构建
Grok 4.7 is our most capable model for coding and knowledge work. It works longer on difficult tasks, checks its own work more carefully, and comes with our best-calibrated safeguards to date. Served at the same price and speed as Grok 4.6, it is highly competitive in its class.
Grok 4.7 是我们功能最强大的编码与知识工作模型。它能在困难任务上运行更长时间，更仔细地检查自身工作，并配备迄今为止校准最佳的 safeguards（安全护栏）。其价格与速度与 Grok 4.6 相同，在同类产品中极具竞争力。
A scatter and line chart comparing Fable 5.1, Opus 5, Grok 4.7, GPT-5.6 Sol, and Sonnet 5 scores against average cost per task.55%CursorBench 4.0 score50%45%40%35%30%25%20%$18$15$12$9$6$3$0Average cost per taskFable 5.1Opus 5GPT-5.6 SolSonnet 5Grok 4.7
散点图和折线图比较了 Fable 5.1、Opus 5、Grok 4.7、GPT-5.6 Sol 和 Sonnet 5 的得分与每项任务的平均成本。55% CursorBench 4.0 得分 50% 45% 40% 35% 30% 25% 20% $18 $15 $12 $9 $6 $3 $0 每项任务的平均成本 Fable 5.1 Opus 5 GPT-5.6 Sol Sonnet 5 Grok 4.7
CostTokensSteps
成本 Token 步骤
On CursorBench 4.0, which stresses longer-running coding tasks, Grok 4.7 is at the frontier in price-performance.
在强调长运行编码任务的 CursorBench 4.0 上，Grok 4.7 在性价比方面处于前沿水平。
Model Improvements
模型改进
Grok 4.7 uses a new, larger base model compared to Grok 4.6. It was trained with a longer reinforcement learning run on a harder mix of tasks, weighted toward problems that take many hours to complete. The model is better at verifying its own work and managing longer context. We also trained Grok 4.7 to natively understand the Grok Bot harness, making it better at conversational tasks and general knowledge work.
与 Grok 4.6 相比，Grok 4.7 使用了新的、更大的基础模型。它经过更长时间的强化学习训练，任务混合难度更高，权重偏向需要数小时才能完成的问题。该模型在验证自身工作和管理更长上下文方面表现更佳。我们还训练 Grok 4.7 原生理解 Grok Bot harness，使其在对话任务和通用知识工作方面表现更好。
Grok 4.7 xHigh
Grok 4.7 xHigh
Grok 4.6 High
Grok 4.6 High
GPT-5.6 Sol Max
GPT-5.6 Sol Max
Fable 5.1 Max
Fable 5.1 Max
Input token price$ per million
输入 token 价格 每百万美元
$2
$2
$2
$2
$4
$4
$10
$10
Output token price$ per million
输出 token 价格 每百万美元
$6
$6
$6
$6
$20
$20
$50
$50
Software engineeringCursorBench 4.0
软件工程 CursorBench 4.0
46.3%
46.3%
40.4%
40.4%
41.7%
41.7%
51.8%
51.8%
Software engineeringDeepSWE v1.1
软件工程 DeepSWE v1.1
71.0%*
71.0%*
65.2%
65.2%
72.7%
72.7%
70.0%
70.0%
Electrical engineeringEEBench
电气工程 EEBench
64.0%
64.0%
53.0%
53.0%
39.4%
39.4%
56.4%
56.4%
Multi-hour office workAA Briefcase v1.1
多小时办公工作 AA Briefcase v1.1
1,657
1,657
1,546
1,546
1,487
1,487
1,678
1,678
Multi-hour terminal workTerminal-Bench 4.0
多小时终端工作 Terminal-Bench 4.0
38.0%
38.0%
20.3%
20.3%
37.3%
37.3%
57.9%
57.9%
Legal workHarvey Legal Agent Benchmark
法律工作 Harvey Legal Agent Benchmark
19.6%
19.6%
15.8%
15.8%
2.5%
2.5%
6.7%
6.7%
Clinical reasoningHealthBench Professional
临床推理 HealthBench Professional
56.7%
56.7%
48.5%
48.5%
60.5%
60.5%
62.1%
62.1%
* high effort
* 高努力程度
Token prices and benchmark scores for Grok 4.7, Grok 4.6, GPT-5.6 Sol, and Fable 5.1. Benchmarks are CursorBench 4.0, DeepSWE v1.1, EEBench, AA Briefcase v1.1, Terminal-Bench 4.0, Harvey Legal Agent Benchmark, and HealthBench Professional. An asterisk on Grok 4.7 DeepSWE marks a high-effort score.
Grok 4.7、Grok 4.6、GPT-5.6 Sol 和 Fable 5.1 的代币价格与基准分数。基准测试包括 CursorBench 4.0、DeepSWE v1.1、EEBench、AA Briefcase v1.1、Terminal-Bench 4.0、Harvey Legal Agent Benchmark 和 HealthBench Professional。Grok 4.7 DeepSWE 旁的星号表示高投入得分。
Grok 4.7 is better at creating documents and presentations. In GDPval and AA Briefcase, AI is asked to work on tasks done by professionals such as lawyers, nurses, and financial analysts. Grok 4.7 improves upon Grok 4.6 on both benchmarks and performs comparably to other frontier models.
Grok 4.7 在创建文档和演示文稿方面表现更佳。在 GDPval 和 AA Briefcase 中，AI 被要求完成由律师、护士和金融分析师等专业人士执行的任务。Grok 4.7 在这两项基准测试中均优于 Grok 4.6，并与其他前沿模型表现相当。
Professional knowledge workMulti-hour office workElectrical engineering
专业知识工作多小时办公工作电气工程
Professional knowledge work
专业知识工作
GDPval
GDPval
050010001500Elo score1735Fable 5.1 (max)1695Grok 4.7 (xhigh)1605Grok 4.6 (high)1542GPT-6 Astra (max)
050010001500Elo 评分1735Fable 5.1（最高）1695Grok 4.7（极高）1605Grok 4.6（高）1542GPT-6 Astra（最高）
GDPval, AA Briefcase, and EEBench scores comparing Grok 4.7 with Grok 4.6, Fable 5.1, and GPT-6 Astra.
GDPval、AA Briefcase 和 EEBench 分数对比了 Grok 4.7 与 Grok 4.6、Fable 5.1 和 GPT-6 Astra 的表现。
Safety & Cybersecurity
安全与网络安全
Grok 4.7 was built with an entirely new safeguard stack. It is the strongest model we’ve tested on refusals and jailbreak resistance. In dual-use domains like cybersecurity and biological work, it leads on both utility for benign tasks and safe refusal on dangerous ones, topping LatchBio’s biosafety benchmark at 62.4%.
Grok 4.7 采用了全新的安全防护栈。它是我们在拒绝响应和抵抗越狱方面测试过的最强模型。在网络安全和生物工作等双重用途领域，它在良性任务的实用性和危险任务的安全拒绝方面均领先，并在 LatchBio 的生物安全基准测试中以 62.4% 的成绩位居榜首。
Grok 4.7 balances strong cyber defense capabilities with low refusal rates for legitimate use. It shows the highest safety on HackerBench v0.3, our benchmark for risky and malicious cyber tasks, allowing only 3.3% of risky dual-use prompts through while rarely blocking legitimate security work. We’ve also started giving select cybersecurity partners invite-only access to Grok 4.7’s red-team capabilities for defense research.
Grok 4.7 在强大的网络防御能力与较低的正常用途拒绝率之间取得了平衡。在我们针对高风险和恶意网络任务的基准测试 HackerBench v0.3 中，它展现出最高的安全性，仅允许 3.3% 的高风险双重用途提示通过，而很少阻止合法的安全工作。我们还开始向部分网络安全合作伙伴提供仅限邀请的访问权限，以使用 Grok 4.7 的红队能力进行防御研究。
Pricing and availability
定价与可用性
Grok 4.7 is available today in Cursor and Grok Build. It is also available through the Grok API, third-party coding harnesses, and model routers and cloud platforms.
Grok 4.7 今日已在 Cursor 和 Grok Build 中上线。它还可通过 Grok API、第三方编码工具包以及模型路由器和云平台获取。
The model is priced starting at $2 per million input tokens and $6 per million output tokens. We also serve a fast variant with twice the output speed at twice the price.
该模型的定价为每百万输入代币 2 美元起，每百万输出代币 6 美元。我们还提供一款速度加倍但价格翻倍的快速变体。
Console
控制台
Create an API key
创建 API 密钥
docs.x.ai
docs.x.ai
Read the docs
阅读文档
Try it in Grok Build for free
在 Grok Build 中免费试用
Get started today at x.ai/build.
立即前往 x.ai/build 开始使用。
$ curl -fsSL https://x.ai/cli/install.sh | bash
$ curl -fsSL https://x.ai/cli/install.sh | bash
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