ModelCap

Current AI model comparison

Exact identities · public evidence

DeepSeek

DeepSeek V4.1 Flash

Position

#20

Index

78.8

Measured

4 public benchmark observations across 4 boards

Comparison side 1

Meta

Muse Wonder

Position

#6

Index

83.0

Estimated

1 reported rows against measured corpus ladders (69%); publisher-corpus-prior over 211 held-out anchors (31%); shrunk 0.2 up toward the measured corpus

Comparison side 2

Output / 1M

$0.60

vs $20.00

Context

1M

vs 1M

Weights

Open weights

vs No linked weights