BREAKING
Moonshot Opens Kimi-K2.7-Code
0T
Total params
0B
Active per token
0K
Context window
Gains Over K2.6
Kimi Code v221.8
Program Bench11
MLS Lite31.5
0$
Input /1M
0$
Output /1M
0%
Fewer tokens
Praise Meets Skepticism
Praise
Top open-source coder
Runs locally for free
Strong token efficiency
Concerns
Benchmarks vs reality gap
Long-context weakness
Safety still unproven
Open Models Closing In
AI NEWS BLITZ
Moonshot AI has released its coding-focused Kimi-K2.7-Code as open weights.