BREAKING
Moonshot AI launches Kimi-K2.7-Code
0
T
total params
0
B
active
0
K
context
0
%
Kimi Code Bench v2
0
%
MLS Bench Lite
0
%
fewer tokens
Kimi Code Bench v2 scores
K2.6
50.9
K2.7-Code
62
Opus 4.8
67.4
GPT-5.5
69
Praise vs caveats
Strengths
●
Clean frontend structure
●
Stable long coding loops
●
Lower cost and latency
Concerns
●
Slow on some hosts
●
Proprietary benchmarks
●
KernelBench-Hard regressions
Open-weight contender in Code Arena
AI NEWS BLITZ
Moonshot AI has released its open-source coding model Kimi-K2.7-Code.