ModelCap

Current AI model comparison

Exact identities · public evidence

Kwaipilot

KAT Coder V2.5 Dev

Position

#37

Index

65.1

Estimated

global-corpus-prior over 186 held-out anchors (23%); launch card against 2 resolved peers on 4 rows (77%); exceeds every named peer on 4 of 4 rows; optimism haircut 0 from cross-lab-probe; shrunk 3.8 toward the measured corpus

Comparison side 1

Xiaomi

MiMo-V2.5

Position

#40

Index

63.8

Measured

2 public benchmark observations across 2 boards

Comparison side 2

Output / 1M

Unavailable

vs $0.28

Context

vs 1M

Weights

Open weights

vs Open weights