1 / 3
Swipe to compare

Grok 4.1 Fast (Reasoning) is the reasoning-enabled configuration of Grok 4.1 Fast, with extended thinking always active. The model produces explicit chain-of-thought traces before its final answer, delivering higher accuracy on complex analytical tasks at the cost of additional latency. It shares the same 2M-token context window and tool-calling capabilities as the standard variant, making it suited for tasks where correctness on difficult problems justifies the extra thinking time.

Author
GrokGrok
Release Date
2025-09-15
Knowledge Cutoff
2025-06-01
License
Proprietary
I/O Format
Context Length
2M / 33K
API I/O (1M)
$0.2 / $0.5
How to Use
Grok SuperGrok or above / API Access
Output Speed
114 tok/s
Arena Overall
1432
Intelligence Index
38.6
Coding Index
30.9
Math Index
89.3
LiveBench
60.1
ForecastBench
60.2
GPQA Diamond
85.3%
HLE
17.6%
MMLU-Pro
85.4%
AIME 2025
89.3%
MATH-500
LB Reasoning
80.2
LB Math
83.7
LB Data Analysis
52.2
LiveCodeBench
82.2%
LB Coding
69.6
LB Agentic
31.7
TAU2
93.3%
TerminalBench
24.2%
SciCode
44.2%
IFBench
52.7%
AA-LCR
0.7
Hallucination (HHEM)
19.2%
Factual Consistency (HHEM)
80.8%
LB Language
74.3
LB Instruction Following
28.2