ModelCap

Current AI model comparison

Exact identities · public evidence

Qwen

Qwen2.5 VL 32B Instruct

Position

#118

Index

40.5

Estimated

publisher-corpus-prior over 194 held-out anchors (18%); launch card against 2 resolved peers on 4 rows (82%); exceeds every named peer on 2 of 4 rows; optimism haircut 0 from cross-lab-probe; shrunk 2.1 up toward the measured corpus

Comparison side 1

Qwen

Qwen3 4B Instruct 2507

Position

#149

Index

32.3

Estimated

publisher-corpus-prior over 194 held-out anchors (19%); launch card against 2 resolved peers on 9 rows (81%); exceeds every named peer on 7 of 9 rows; no cross-lab optimism probe available; shrunk 4.1 up toward the measured corpus

Comparison side 2

Output / 1M

Unavailable

vs Unavailable

Context

vs 262K

Weights

Open weights

vs Open weights