Beta
Classificação/Claude Sonnet 4.5
Anthropic

Claude Sonnet 4.5

por Anthropic · Lançado em 2025-09-29

40.6
pontuação média
$3.00/1M
Preço de entrada
$15.00/1M
Preço de saída
1.0M tokens (~500 books)
Janela de contexto
multimodal
Tipo

Tested on 18 benchmarks with 40.6% average. Top scores: MATH level 5 (97.7%), OTIS Mock AIME 2024-2025 (77.8%), GPQA diamond (76.4%).

Pontuações de benchmark

BenchmarkCategoriaPontuaçãoBar
MATH level 5math97.7
OTIS Mock AIME 2024-2025math77.8
GPQA diamondknowledge76.4
SWE-Bench Verified (Bash Only)coding70.6
OSWorldagentic62.9
Cybenchcoding55.0
DeepResearch Benchknowledge52.6
WeirdMLcoding47.7
SimpleBenchreasoning45.2
Terminal Benchcoding42.8
SimpleQA Verifiedknowledge23.6
FrontierMath-2025-02-28-Privatemath15.2
GSO-Benchcoding14.7
ARC-AGI-2reasoning13.6
Chess Puzzlesknowledge12.0
VPCTknowledge9.7
HLEknowledge9.4
FrontierMath-Tier-4-2025-07-01-Privatemath4.2

Modelos similares

GoogleGemini 1.5 Flash (Sep 2024)
Google
40.4
M
Kimi K2.5
moonshotai
40.9
OpenAIGPT-5 Pro
OpenAI
40.9
AlibabaQwen2.5-Coder (1.5B)
Alibaba
40.3

Anthropic Claude Sonnet Cronologia

Claude 3 Sonnet
28.3
N/AN/Actx6 benchmarks
Claude 3.5 Sonnet
28.8+0.5
$6.00/M in200Kctx10 benchmarks
Claude 3.7 Sonnet
42.1+13.3
$3.00/M in(-3)200Kctx20 benchmarks
Claude 3.7 Sonnet (thinking)
42.1
$3.00/M in200Kctx20 benchmarks
Claude Sonnet 4
36.9-5.2
$3.00/M in200Kctx20 benchmarks
Claude Sonnet 4.5
40.6+3.7
$3.00/M in1.0Mctx(+800K)18 benchmarks
Claude Sonnet 4.6
47.3+6.7
$3.00/M in1.0Mctx8 benchmarks