返回开发者专区官方定价 官方定价 官方定价
xAI 模型
探索 xAI 的所有 3 个模型，包括详细定价、优缺点和开发者推荐。
3
模型
$0.200
最低输入价格
2M
最大上下文
2
质量层级
快速推荐
最佳性价比: Grok 4.1 Fast ($0.200/1M)
最佳质量: Grok 4.3
Grok 4.3
FlagshipGeneral-purpose, max capability
适用场景: Best xAI model for general-purpose applications requiring real-time knowledge access.
核心升级
- ◆1M context window — 8x increase over Grok 3's 128K
- ◆Price: $1.25/$2.50 — 60% cheaper than Grok 3 ($3/$15)
- ◆Real-time web + X (Twitter) search — unique live knowledge access
- ◆Context caching: $0.20/M — 84% savings for repeated prefixes
- ◆32K max output — 4x Grok 3's 8K for long-form generation
输入价格
$1.25
per 1M tokens
输出价格
$2.50
per 1M tokens
缓存输入
$0.200
per 1M tokens
批量输入
—
per 1M tokens
上下文窗口: 1M
最大输出: 32,000 tokens
知识截止日期: 2025-06
视觉函数调用微调JSON 模式
优点
- 1M context window at $1.25/M input
- Most intelligent xAI model
- Real-time web and X search
缺点
- No batch API
- No fine-tuning
- Newer ecosystem
性能
输出速度~65 tok/s
速率限制3,000 RPM
多模态能力
图像输入图像输出音频输入音频输出
基准测试
MMLU
87.5%
SWE-bench Verified
62.0%
GPQA
73.0%
Grok 4.1 Fast
Mid-tierCost-optimized production workloads
适用场景: Best for latency-sensitive production apps and long-document processing on a budget.
核心升级
- ◆2M context window — largest among all mid-tier models
- ◆$0.20/M input — 6x cheaper than Grok 4.3 for high-volume tasks
- ◆Ultra-low latency — optimized for sub-500ms response times
- ◆16K max output — 2x Grok 3's 8K for longer generations
- ◆Context caching: $0.05/M — 75% savings for repeated prefixes
输入价格
$0.200
per 1M tokens
输出价格
$0.500
per 1M tokens
缓存输入
$0.050
per 1M tokens
批量输入
—
per 1M tokens
上下文窗口: 2M
最大输出: 16,000 tokens
知识截止日期: 2025-04
视觉函数调用微调JSON 模式
优点
- 2M context window — largest among mid-tier
- Extremely fast response times
- One of lowest rates for frontier APIs
缺点
- Lower quality than Grok 4.3
- No batch API
- No fine-tuning
性能
输出速度~110 tok/s
速率限制8,000 RPM
多模态能力
图像输入图像输出音频输入音频输出
基准测试
MMLU
83.0%
HumanEval
80.0%
Grok 3
FlagshipReal-time info, analysis
适用场景: When real-time X/Twitter data or very current knowledge is essential.
核心升级
- ◆Real-time X/Twitter integration — unique live social media knowledge
- ◆Knowledge cutoff: 2025-02 — one of the most current available
- ◆Vision + function calling — full multimodal capability at launch
- ◆No caching or batch API — upgrade to Grok 4.3 for caching support
- ◆128K context — superseded by Grok 4.3's 1M for long-document tasks
输入价格
$3.00
per 1M tokens
输出价格
$15.00
per 1M tokens
缓存输入
—
per 1M tokens
批量输入
—
per 1M tokens
上下文窗口: 131K
最大输出: 8,192 tokens
知识截止日期: 2025-02
视觉函数调用微调JSON 模式
优点
- Very recent knowledge cutoff
- Real-time X/Twitter integration
- Vision + function calling
缺点
- Expensive ($3/$15)
- No caching or batch
- No fine-tuning
性能
输出速度~55 tok/s
速率限制2,000 RPM
多模态能力
图像输入图像输出音频输入音频输出
基准测试
MMLU
85.0%
GPQA
68.0%
MATH
70.0%
并排比较
| 模型 | 层级 | 输入 | 输出 | 上下文 |
|---|---|---|---|---|
| Grok 4.3 | Flagship | $1.25 | $2.50 | 1M |
| Grok 4.1 Fast | Mid-tier | $0.200 | $0.500 | 2M |
| Grok 3 | Flagship | $3.00 | $15.00 | 131K |