Grok 4

xAI

Frontier ModelReasoningProprietary

Context

Release Date: Jul 10, 2025
Window: 256k

PricingPer 1M tokens

Input: $3
Output: $15
Blended 3:1: $6

Capabilities

Speed: 75 t/s
Input
Output
Reasoning tokens

Latency

TTFT: 5.92 ms
500 token response: 12.50 s

Benchmarks

Reasoning: ●●●●○
MMLU Pro: 87.0%
GPQA: 88.0%
HLE: 38.6%
AIME: 94.0%
LiveCodeBench: 79.3%
HumanEval: 98.0%

Grok 4

Grok 4 is xAI's flagship LLM with a 256k window, top reasoning benchmarks, and moderately fast 75 token-per-second output.

Grok 4 is an RL-optimized next variant of the Grok 3 base model.