Claude Sonnet 4.5 (Thinking)

Release Date: 9/29/2025

Benchmarked by
Vals Index

Claude Sonnet 4.5 is Anthropic's best model for complex agents and coding, supporting 200K context (1M tokens beta), with state-of-the-art reasoning, intelligence, and tool integration. It excels in autonomous agent workflows and computer use.

Avg. Accuracy

74.3%

Latency

170.6s

Cost (In/Out)

3.00 / 15.00

Context Window

1000k

Max Output Tokens

64k

Input Modality

Benchmarks

Accuracy

Rankings

SAGE

0.0%

21/ 21

FinanceAgent

0.0%

42/ 42

CorpFin

0.0%

55/ 55

CaseLaw

0.0%

38/ 38

TaxEval

0.0%

72/ 72

MortgageTax

0.0%

44/ 44

AIME

0.0%

62/ 62

MGSM

0.0%

65/ 65

LegalBench

0.0%

87/ 87

MedQA

0.0%

68/ 68

GPQA

0.0%

64/ 64

MMLU Pro

0.0%

62/ 62

MMMU

0.0%

41/ 41

LiveCodeBench

0.0%

61/ 61

Terminal-Bench

0.0%

27/ 27

SWE-bench

0.0%

23/ 23

Vals Index

0.0%

17/ 17

Academic Benchmarks
Proprietary Benchmarks (contact us to get access)

Join our mailing list to receive benchmark updates

Model benchmarks are seriously lacking. With Vals AI, we report how language models perform on the industry-specific tasks where they will be used.

By subscribing, I agree to Vals' Privacy Policy.