ModelCap
Current AI model comparison
Exact identities · public evidence
Microsoft
Phi mini MoE instruct
Position
#126
Index
44.1
Estimated
global-corpus-prior over 209 held-out anchors (68%); launch card against 4 resolved peers on 3 rows (32%); exceeds every named peer on 3 of 3 rows; optimism haircut 6.91 from cross-lab-probe; shrunk 26.9 up toward the measured corpus
Comparison side 1
Qwen
Qwen3.5 4B
Position
#124
Index
44.6
Estimated
publisher-corpus-prior over 209 held-out anchors (22%); launch card against 3 resolved peers on 10 rows (78%); exceeds every named peer on 2 of 10 rows; optimism haircut 0.6 from cross-lab-probe; shrunk 3.2 up toward the measured corpus
Comparison side 2
Output / 1M
Unavailable
vs Unavailable
Context
—
vs —
Weights
Open weights
vs Open weights