跳轉到主要內容

Grok 4

SpaceXAIGrokProprietary

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
—
支援模態
image, text

能力雷達圖

42
general
82
coding
94
reasoning
69
science
90
agents
80
multimodal

排行榜排名

領域#排名分數來源
程式碼能力榜150
74.0
AA
通用能力榜152
60.0
AA
科學能力151
66.0
AA

基準測試分數 (LLM Stats)

(LLM Stats (zeroeval))

Math

AIME 202591.7%自報
HMMT2590.0%自報
USAMO2537.5%自報

Reasoning

GPQANYU + Cohere + Anthropic (2023)87.5%自報
LiveCodeBench79.0%自報
Humanity's Last Exam40.0%自報
ARC-AGI v215.9%自報

AA 評測指數

(Artificial Analysis)
Math 500(OpenAI (2024), subset of Hendrycks et al. MATH (2021))
99.0
Aime(MAA (Mathematical Association of America))
94.3
Math Index(Artificial Analysis)
92.7
Aime 25(MAA (Mathematical Association of America))
92.7
Gpqa(NYU + Cohere + Anthropic (2023))
87.7
Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))
86.6
Livecodebench(UC Berkeley + MIT + Cornell (2024))
81.9
Tau2(Sierra + U Toronto + Vector Institute (2025))
74.9
Lcr(Artificial Analysis)
68.0
Ifbench(Google Research (2023))
53.7
Terminalbench Hard(Stanford × Laude Institute (2026))
37.9
Hle(Center for AI Safety + Scale AI (2025))
26.7
Intelligence Index(Artificial Analysis)
22.5

LLM Stats 分類評分

(LLM Stats (zeroeval))
Physics
90
Biology
90
Chemistry
90
Code
80
Math
60
Reasoning
60
General
60
Vision
30
Spatial Reasoning
20

定價

輸入價格$3 / 1M tokens
輸出價格$15 / 1M tokens
混合價格(3:1)$6 / 1M tokens

速度

Tokens/秒0.0
首Token延遲0.00s
首回答延遲0.00s

供應商價格排行

供應商價格排行

3 個供應商

最便宜: SpaceXAI最貴: DevPass (LLM Gateway)
供應商輸入輸出
1SpaceXAI主要
$3
$15
2Helicone
$3
$15
3DevPass (LLM Gateway)
$3
$15

比較該模型在不同 API 供應商之間的定價。

外部連結