ModelCap
Current AI model comparison
Exact identities · public evidence
DeepSeek
DeepSeek V4 Flash Vision Exp
Position
#23
Index
71.2
Estimated
publisher-corpus-prior over 182 held-out anchors (13%); launch card against 2 resolved peers on 7 rows (87%); exceeds every named peer on 1 of 7 rows; no cross-lab optimism probe available; shrunk 2 toward the measured corpus
Comparison side 1
DeepSeek
DeepSeek V4 Pro 0813
Position
#20
Index
73.6
Measured
4 public benchmark observations across 4 boards
Comparison side 2
Output / 1M
$0.66
vs $1.74
Context
1M
vs 1M
Weights
Open weights
vs Open weights