BREAKING
Kimi K3: 2.8T-Parameter Open Model
0T
parameters
0
active experts
0M
token context
GDPval-AA v2: K3 Ranks Third
Claude Fable 5 Max1815
GPT-5.6 Sol Max1748
Kimi K31687
0$
per M input
0$
per M cached
0$
per M output
Strengths vs Limitations
Strengths
Top Frontend Code Arena score
Agentic and long-document work
Strong price-to-performance
Limitations
Trails Claude on complex tasks
Verbose, higher token costs
Elevated tool-call failures
Weights Follow July 27
AI NEWS BLITZ
Moonshot AI has released Kimi K3, one of the largest open-weight models ever made public.