ModelBeat
Models/Claude Sonnet 4.5
All models
A

Anthropic: Claude Sonnet 4.5

anthropic/claude-sonnet-4.5

Claude Sonnet 4.5 is Anthropic’s most advanced Sonnet model to date, optimized for real-world agents and coding workflows.

Context
1M
Input
$3 / 1M
Output
$15 / 1M
Released
Sep 29, 2025
Modalities
textimagefiletext

Benchmarks

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

37.0
Intelligence Index
Epoch AI
37th percentile of tracked models
38.0
Coding Index
Epoch AI
38th percentile of tracked models
45.0
Agentic Index
Epoch AI
45th percentile of tracked models

Reasoning

7 evals
ARC-AGI63.7%

Abstract visual reasoning

ARC-AGI-213.6%

Harder abstract reasoning

GPQA Diamond82.3%

Graduate-level scientific reasoning

Humanity's Last Exam13.7%

Frontier of human expert knowledge

SimpleBench54.3%

Common-sense trick questions

SimpleQA Verified23.6%

Factual accuracy & hallucination

WeirdML47.7%

Novel ML problem-solving

Coding

3 evals
GSO (code optimization)14.7%

Code performance optimization

SWE-bench Verified71.3%

Real GitHub issue resolution

WebDev Arena1391

Human-rated web development

Math

4 evals
AIME 2024/202577.8%

Olympiad-qualifier math

FrontierMath15.2%

Research-level math problems

FrontierMath Tier 44.2%

Hardest research math

MATH Level 597.7%

Hardest competition math

Agentic & Tools

3 evals
GDPval (win/tie rate)50.3%

Economically valuable work

METR task horizon2.0 h

Autonomous task length

Terminal-Bench46.5%

Command-line agentic tasks

In the news

3

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