Grok 4.1 Fast is xAI's agentic tool-calling model, optimized for real-world use cases such as customer support and deep research. It features a 2M-token context window — the largest among Western frontier models — and focuses on significantly reduced hallucination rates for information-seeking tasks. Reasoning can be enabled or disabled via the API's reasoning parameter, allowing developers to choose between speed-optimized direct answers and deeper analytical responses.
xAI SuperGrokxAI SuperGrok HeavyAPI|VisionWeb SearchFile|Proprietary Model
Knowledge Cutoff
2024-11
Input → Output Format
Context Memory
2MIN30KOUT
AI Performance Evaluation
Arena Overall Score
1432
±4As of 2026-05-01
Overall Rank
No.63
48,702 Votes
Arena by Ability
Hard Prompts
1442±4No.77
Expert Knowledge
1441±11No.79
Instruction Following
1401±6No.94
Conversation Memory
1417±7No.87
Creative
1410±7No.53
Coding
1465±6No.81
Math
1423±11No.72
Arena by Occupation
Creative Writing
1403±6No.75
Social Sciences
1450±7No.62
Media
1400±7No.62
Business
1417±7No.76
Healthcare
1447±11No.72
Legal
1427±11No.80
Software
1460±5No.70
Mathematics
1423±13No.76
Source:Arena Intelligence
Overall
AA Intelligence Index
24%↓16%
LiveBench
32%↓29%
ForecastBench
56%↓4%
Reasoning & Math
AA Math Index
34%↓40%
GPQA Diamond
64%↓18%
HLE
5.0%↓13%
MMLU-Pro
74%↓7%
AIME 2025
34%↓40%
LB Reasoning
23%↓46%
LB Math
39%↓35%
LB Data
41%↓13%
Coding
AA Coding Index
20%↓17%
LiveCodeBench
40%↓26%
LB Coding
54%↓19%
LB Agentic
10%↓35%
TAU2
64%↓17%
TerminalBench
14%↓20%
SciCode
30%↓12%
Language & Instructions
IFBench
37%↓27%
AA-LCR
22%↓40%
Hallucination (HHEM)
18%↑8%
Factual (HHEM)
82%↓8%
LB Language
50%↓22%
LB IF
17%↓34%
Output Speed
Standard Mode
79tok/s↑2
First Output 0.45s
Reasoning Mode
94tok/s↑7
First Output 11.80s
