BENCHMARK REGISTRY
Public Artificial Intelligence Benchmark Records Database

Claude 3 Haiku

MODEL RECORD: ANTHR-H3

Organization
Anthropic
Family
Claude
Model number
3
Tier
H
Status
ACTIVE
Release date
2024-03-04

Table may be scrolled horizontally on narrow displays.

Record No. Model Benchmark Metric Score Eval Date Source Model ID
BR-002H3-001 Claude 3 Haiku Artificial Analysis Intelligence Index Overall 3.9 Unknown Primary source ANTHR-H3
BR-002H3-002 Claude 3 Haiku GPQA Diamond Accuracy 37.4% Unknown Primary source ANTHR-H3
BR-002H3-003 Claude 3 Haiku Humanity's Last Exam No tools Overall 3.9% Unknown Primary source ANTHR-H3
BR-002H3-004 Claude 3 Haiku SciCode Overall 18.6% Unknown Primary source ANTHR-H3
BR-002H3-005 Claude 3 Haiku tau2-bench Overall 21.1% Unknown Primary source ANTHR-H3
BR-002H3-006 Claude 3 Haiku Terminal-Bench Hard Overall 0.8% Unknown Primary source ANTHR-H3
BR-002H3-007 Claude 3 Haiku CritPt Overall 0% Unknown Primary source ANTHR-H3
BR-002H3-008 Claude 3 Haiku AA-Omniscience Overall -47.6 Unknown Primary source ANTHR-H3
BR-002H3-009 Claude 3 Haiku AA-Omniscience Accuracy 17.2% Unknown Primary source ANTHR-H3
BR-002H3-010 Claude 3 Haiku IFBench Overall 36.1% Unknown Primary source ANTHR-H3
BR-002H3-011 Claude 3 Haiku MMMU-Pro Overall 30.8% Unknown Primary source ANTHR-H3
BR-002H3-012 Claude 3 Haiku AA-LCR Overall 21% Unknown Primary source ANTHR-H3