Compare

grok-3

Comparing 1 model

 Grok 3
Overview
StatusActive
Released2/18/2025
SummaryxAI's frontier reasoning model with real-time web access.
Pricing (per 1M tokens)
Input$3/1M
Output$15/1M
Context and limits
Context window131K
Max output16K
Modalities
Supportedtext, image
Capabilities
Tool callingYes
Structured outputsYes
Open sourceNo
License
Benchmarks
Aider Polyglot
% pass@2
53.3
Third-party
AIME 2024
% accuracy
83.9
Self-reported
GPQA Diamond
% accuracy
84.6
Self-reported
HumanEval
pass@1 %
88.5
Self-reported
MATH
% accuracy
93.3
Self-reported
MMLU
% accuracy
87
Self-reported
MMLU-Pro
% accuracy
79.3
Self-reported
SWE-bench Verified
% resolved
50
Third-party

Presets