跳轉到主要內容

Grok 4

xAIGrokProprietary

描述

Grok 4, announced by xAI in summer 2025, represents a major leap in AI capabilities, described as 'the smartest AI in the world.' Built on version 6 of xAI's foundation model, it uses 100x more training compute than Grok 2 and 10x more reinforcement learning compute than Grok 3. The model achieves PhD-level performance across all academic disciplines simultaneously, scoring perfect on standardized tests like the SAT and near-perfect on graduate exams like the GRE. Unlike Grok 3, tool usage is built into the training process rather than relying on generalization. Trained using 200,000 GPUs, Grok 4 excels at complex reasoning, mathematical problem-solving, and coding tasks, though it has acknowledged weaknesses in multimodal capabilities that are being addressed in the next version.

發布日期
2025-07-10
參數規模
上下文長度
256K
支援模態
file, image, text

能力雷達圖

52
general
56
coding
94
reasoning
60
science估算
0
agents
80
multimodal

Science 在缺少專門科學評測時使用推理能力代理估算。

排行榜排名

領域#排名分數來源
代码能力榜37
76.0
AA
通用能力榜77
73.0
AA
数学推理11
96.0
AA
推理能力103
16.0
LS
科学能力43
74.0
AA

基準測試分數 (LLM Stats)

Biology

GPQA87.5%自報

Code

LiveCodeBench79.0%自報

Math

AIME 202591.7%自報
HMMT2590.0%自報
Humanity's Last Exam40.0%自報
USAMO2537.5%自報

Reasoning

ARC-AGI v215.9%自報

AA 評測指數

Math Index
92.7
Intelligence Index
41.5
Coding Index
40.5
Math 500
1.0
Aime
0.9
Aime 25
0.9
Gpqa
0.9
Mmlu Pro
0.9
Livecodebench
0.8
Tau2
0.7
Lcr
0.7
Ifbench
0.5
Scicode
0.5
Terminalbench Hard
0.4
Hle
0.2

LLM Stats 分類評分

Biology
90
Chemistry
90
Physics
90
Code
80
General
80
Math
60
Reasoning
60
Vision
30
Spatial Reasoning
20

定價

輸入價格$4.25 / 1M tokens
輸出價格$21.25 / 1M tokens
混合價格(3:1)$8.5 / 1M tokens

速度

Tokens/秒47.2 tokens/s
首Token延遲14.13s
首回答延遲14.13s

可用提供商

(LS 內部計價單位)

暫無提供商資料

外部連結