ModelCap

Current AI model comparison

Exact identities · public evidence

DeepSeek

DeepSeek V3.2

Position

#54

Index

58.1

Measured

4 public benchmark observations across 4 boards

Comparison side 1

Stepfun Ai

Step3 VL 10B

Position

#50

Index

60.0

Estimated

global-corpus-prior over 183 held-out anchors (17%); launch card against 2 resolved peers on 4 rows (83%); exceeds every named peer on 3 of 4 rows; no cross-lab optimism probe available; shrunk 1.4 toward the measured corpus

Comparison side 2

Output / 1M

$0.40

vs Unavailable

Context

164K

vs

Weights

Open weights

vs Open weights