ModelCap
Current AI model comparison
Exact identities · public evidence
Anthropic
Claude Haiku 5.5
Position
#59
Index
63.5
Measured
1 reported rows against measured corpus ladders (24%); publisher-corpus-prior over 213 held-out anchors (14%); measured predecessor anthropic/claude-haiku-4.5 less the succession penalty (51%); 1 specialist-board observation at 2.5% support (11%); shrunk 2.3 up toward the measured corpus
Comparison side 1
Anthropic
Claude Opus 5.5
Position
#3
Index
86.6
Measured
4 public benchmark observations across 4 boards
Comparison side 2
Output / 1M
$0.50
vs $20.00
Context
1M
vs 1M
Weights
No linked weights
vs No linked weights