#24
Grok 3
grok-3

Grok 3 is a large language model from xAI focused on strong step-by-step reasoning, math, and coding while still acting as a capable general assistant. It was trained with substantial reinforcement learning to support longer chains of thought, use tools, and draw on web search for grounded answers across a wide context window. Day-to-day it provides solidly helpful explanations, emotionally aware dialogue, and reliable instruction following, though its creativity and overall quality reflect an earlier generation of the Grok family compared with more modern systems.

Performance Metrics

Helpfulness
Instruction Following
Comprehension
Empathy
Creative Writing
82.40
Helpfulness
83.5
Empathy
82
Instruction Following
86
Creative Writing
76
Comprehension
84.5
Speed
Avg 68 tok/s

Model Specifications

Release Date
February 19, 2025
Lab
xAI
Type
Proprietary
Context Size
131.1K
Max Output Tokens
8.2K
Cost per 1M tokens
$3.00 / $15.00
Model Inputs
Text, Images, Code, Web
Model Outputs
Text
Tool Calling
Enabled

Compare With Similar Models