ModelCap

Current AI model comparison

Exact identities · public evidence

Microsoft

Phi 3.5 mini instruct

Position

#221

Index

11.5

Estimated

publisher-corpus-prior over 215 held-out anchors (26%); launch card against 4 resolved peers on 6 rows (74%); exceeds every named peer on 1 of 6 rows; optimism haircut 0.67 from cross-lab-probe; shrunk 1.6 toward the measured corpus

Comparison side 1

Mistral AI

Mixtral 8x22B Instruct

Position

#219

Index

12.5

Measured

2 public benchmark observations across 2 boards

Comparison side 2

Output / 1M

Unavailable

vs $6.00

Context

vs 66K

Weights

Open weights

vs Open weights