K
K2 Think V2
Last updated: Recently
Model Type
Text
Max Context
-
Max Output
-
Parameter Scale
-
K2 Think V2 by MBZUAI Institute of Foundation Models: Text model.
Input Modality
-
Output Modality
-
Inference Speed
-
Latest Release Date
12/15/2025
SDK Ecosystem
-
artificial-analysismanufactureraa-bootstrap
Model Overview Fields
name
K2 Think V2
Release date
12/15/2025
Performance
-
model_creator.name
MBZUAI Institute of Foundation Models
AA Evaluation Scores
AA Intelligence Index
24.1
AA Coding Index
15.5
AA Math Index
-
MMLU Pro
-
GPQA
0.713
HLE
0.095
LiveCodeBench
-
SciCode
0.33
Math-500
-
AIME
-
AIME 2025
-
IFBench
0.628
LCR
0.527
TerminalBench Hard
0.068
TAU2
0.254
Output Speed (tokens/s)
-
TTFT (s)
-
First Answer Token (s)
-

