1 / 3
Swipe to compare

Grok 4.20 (Reasoning) is the reasoning-enabled configuration of xAI's Grok 4.20, utilizing extended internal thinking to work through problems before presenting answers. Combined with the model's native multi-agent architecture and cross-agent verification, it delivers the highest accuracy in the Grok lineup on tasks requiring deep logic, mathematical reasoning, and complex multi-step problem solving. It supports the same 2M-token context window, strict prompt adherence, and the industry's lowest hallucination rate among its class.

Author
GrokGrok
Release Date
2026-03-31
Knowledge Cutoff
Unknown
License
Proprietary
I/O Format
Context Length
2M / 2M
API I/O (1M)
$2 / $6
How to Use
Grok SuperGrok Heavy or above / API Access
Output Speed
113 tok/s
Arena Overall
1482
Intelligence Index
49.3
Coding Index
40.5
Math Index
LiveBench
69.0
ForecastBench
GPQA Diamond
91.1%
HLE
32.2%
MMLU-Pro
AIME 2025
MATH-500
LB Reasoning
75.3
LB Math
87.1
LB Data Analysis
62.9
LiveCodeBench
LB Coding
66.1
LB Agentic
43.3
TAU2
93.0%
TerminalBench
37.9%
SciCode
45.6%
IFBench
81.2%
AA-LCR
0.6
Hallucination (HHEM)
Factual Consistency (HHEM)
LB Language
77.7
LB Instruction Following
63.4