ModelBeat
Models/Claude Haiku 4.5
All models
A

Anthropic: Claude Haiku 4.5

anthropic/claude-haiku-4.5

Claude Haiku 4.5 is Anthropic’s fastest and most efficient model, delivering near-frontier intelligence at a fraction of the cost and latency of larger Claude models.

Context
200K
Input
$1 / 1M
Output
$5 / 1M
Released
Oct 15, 2025
Modalities
textimagefiletext

Benchmarks

Scores on standardized evaluations. Higher is better — and rank shows where Claude Haiku 4.5 lands among all models tracked on Model Beat.

20.0
Intelligence Index
Epoch AI
20th percentile of tracked models
24.0
Coding Index
Epoch AI
24th percentile of tracked models
26.0
Agentic Index
Epoch AI
26th percentile of tracked models

Reasoning

5 evals
ARC-AGI47.7%

Abstract visual reasoning

ARC-AGI-24.0%

Harder abstract reasoning

GPQA Diamond71.2%

Graduate-level scientific reasoning

SimpleQA Verified5.9%

Factual accuracy & hallucination

WeirdML45.4%

Novel ML problem-solving

Coding

1 evals
WebDev Arena1322

Human-rated web development

Math

4 evals
AIME 2024/202566.7%

Olympiad-qualifier math

FrontierMath5.9%

Research-level math problems

FrontierMath Tier 42.1%

Hardest research math

MATH Level 596.4%

Hardest competition math

Agentic & Tools

2 evals
APEX8.9%

Multi-step agentic tasks

Terminal-Bench35.5%

Command-line agentic tasks

In the news

1

Model & benchmark data from Epoch AI (CC BY); pricing, specs & descriptions from OpenRouter.