BREAKING
Moonshot AI Launches Kimi K3
0
T
parameters
0
M
token context
From K2 to K3
1
K2 open-weight MoE
↓
2
K2.5 / K2.6
↓
3
K2.7 Code
↓
4
K3 flagship
Strengths vs caveats
Strengths
●
Coding and agentic performance
●
Agent Swarm parallel tasks
●
Strong price-to-performance
Caveats
●
Early-testing benchmarks
●
Web search discouraged
●
Vision input restrictions
Open weights still unconfirmed
AI NEWS BLITZ
AI NEWS BLITZ
Moonshot AI has released Kimi K3, its most capable model yet.