Back to Leaderboard

Claude Haiku 4.5

Anthropic
Anthropic200K context$8.15/1K pages2025-10-29
Overall Rank
#16
of 27 models
Overall Score
70.2
avg across benchmarks
Best Task
Key Information Extraction
85.6
Weakest Task
Visual QA
59.2

Benchmark Performance

OlmOCR Benchv1.0
22/27
OverallArXiv MathH&FLong/TinyMulti-ColOld ScansScans MathTables
58.255.437.636.262.819.679.583.4
OmniDocBenchv1.5
15/27
OverallText Edit↓CDM↑TEDS↑TEDS-S↑Read Order↓
79.60.22484.277.183.80.178
IDP Core Benchv1.0
16/27
OverallKIEOCRTableVQA
72.985.665.081.759.2

Capability Profile

Strength Analysis

Auto-generated from benchmark scores

Strengths

  • Key Information Extraction85.6
  • Formula84.2

Weaknesses

  • Visual QA59.2
  • Layout & Order72.5