Moonshot AI

Kimi K2-Instruct-0905

Zero-eval
#1Multi-Challenge
#2AutoLogi
#2ACEBench
+3 more

by Moonshot AI

+
+
+
+
About

Kimi K2 Instruct-0905 represents a specific release iteration of the K2 Instruct model, developed to incorporate refinements and improvements. Built to provide enhanced instruction-following based on deployment feedback, it continues the evolution of Moonshot's instruction-tuned offerings.

+
+
+
+
Timeline
AnnouncedSep 5, 2025
ReleasedSep 5, 2025
+
+
+
+
Specifications
Training Tokens15.5T
+
+
+
+
License & Family
License
MIT
Performance Overview
Performance metrics and category breakdown

Overall Performance

29 benchmarks
Average Score
64.1%
Best Score
97.4%
High Performers (80%+)
8
+
+
+
+
All Benchmark Results for Kimi K2-Instruct-0905
Complete list of benchmark scores with detailed information
MATH-500
text
0.97
97.4%
Self-reported
MMLU-Redux
text
0.93
92.7%
Self-reported
IFEval
text
0.90
89.8%
Self-reported
MMLU
text
0.90
89.5%
Self-reported
AutoLogi
text
0.90
89.5%
Self-reported
ZebraLogic
text
0.89
89.0%
Self-reported
MultiPL-E
text
0.86
85.7%
Self-reported
MMLU-Pro
text
0.81
81.1%
Self-reported
ACEBench
text
0.77
76.5%
Self-reported
LiveBench
text
0.76
76.4%
Self-reported
Showing 1 to 10 of 29 benchmarks
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+