ModelCap

Current AI model comparison

Exact identities · public evidence

IBM Granite

granite 3.3 8b instruct

Position

#225

Index

10.4

Estimated

publisher-corpus-prior over 215 held-out anchors (8%); opens above measured predecessor ibm-granite/granite-3.1-8b-instruct on same-family ordering evidence (30%); launch card against 3 resolved peers on 5 rows (62%); exceeds every named peer on 2 of 5 rows; optimism haircut 1.61 from cross-lab-probe; shrunk 0.8 up toward the measured corpus

Comparison side 1

Microsoft

MediPhi PubMed

Position

#231

Index

10.3

Inherited

finetune of microsoft/Phi-3.5-mini-instruct · finetune (100%)

Comparison side 2

Output / 1M

Unavailable

vs Unavailable

Context

vs

Weights

Open weights

vs Open weights