BREAKING
MiniMax M3 leads open-model test
Quality scores: M3 and GLM-5.2 tie
M3
8
GLM-5.2
8
DeepSeek V4
7
Kimi K2.7
5
0
$
MiniMax M3
0
$
GLM-5.2
0
x
cost gap
Different strengths per model
MiniMax M3
~95 tok/s
●
Cheap and fast
●
Agentic and long-context
GLM-5.2
Index 51
●
Top intelligence index
●
Strong planning
One task, many caveats
Open-model price race continues
AI NEWS BLITZ
Four open-weight models faced the same design task, and MiniMax M3 stood out on cost.