NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/Claude Sonnet 4
Anthropic logo

Claude Sonnet 4

by Anthropic · Released May 2025

Multimodal
44.5
avg score
Rank #176
Compare
Better than 44% of all models
Context
200K tokens (~100 books)
Input $/1M
$3.00
Output $/1M
$15.00
Type
multimodal
License
Proprietary
Benchmarks
32 tested
Data as of
About

Claude Sonnet 4 significantly enhances the capabilities of its predecessor, Sonnet 3.7, excelling in both coding and reasoning tasks with improved precision and controllability. Achieving state-of-the-art performance on SWE-bench (72.7%),...

Tested on 32 benchmarks · BenchGecko score 44.5. Top scores: MASK (95.3%), OpenCompass — IFEval (88.3%), MATH level 5 (84.4%).

Looking for similar performance at lower cost?
Qwen3.7 Flash scores 44.6 (100% as good) at $0.03/1M input · 99% cheaper
Capabilities
coding
43.3
#126 globally
reasoning
26.8
#132 globally
math
46.3
#132 globally
knowledge
45.6
#160 globally
agentic
28.8
#39 globally
general
52.6
#26 globally
safety
95.3
#3 globally
language
88.3
#22 globally
Benchmark Scores
Compare All
Tested on 32 benchmarks · Ranked across 8 categories
Score Distribution (all 312 models)
0255075100
▲ You are here
codingCompare coding →
SWE-Bench Verified (Bash Only)

SWE-bench Verified solved using only bash commands, no specialized frameworks. Tests raw terminal-based problem solving.

64.9·
Aider polyglot

Multi-language code editing from Aider. Tests editing ability across Python, JavaScript, TypeScript, Java, C++, Go, Rust, and more.

61.3·
OpenCompass — LiveCodeBenchV6

OpenCompass Live Code Bench v6. Fresh competitive programming problems to evaluate code generation without memorization.

47.5·
reasoningCompare reasoning →
ARC-AGI

Abstraction and Reasoning Corpus. Tests fluid intelligence through novel visual pattern recognition puzzles. Core measure of general intelligence.

40.0·
SimpleBench

Deceptively simple questions that humans find easy but AI models often get wrong. Tests common sense and reasoning gaps.

34.6·
ARC-AGI-2

ARC-AGI 2, harder sequel to ARC. More complex abstract reasoning patterns that test generalization ability beyond training data.

5.9·
mathCompare math →
MATH level 5

Competition-level math from AMC, AIME, and olympiad problems. Level 5 is the hardest tier, requiring creative problem-solving.

84.4·
OTIS Mock AIME 2024-2025

Mock AIME (American Invitational Mathematics Exam) problems from OTIS. Tests mathematical competition performance.

71.1·
OpenCompass — AIME2025

OpenCompass evaluation on AIME 2025 problems. Tests mathematical reasoning on fresh competition problems.

68.7·
Quick compare:Alibaba Qwen logovs Qwen3.7 FlashAnthropic logovs Claude Sonnet 4.5Google DeepMind logovs Gemini 2.0 Flash
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Model Family · Anthropic Claude Sonnet
Jan 2024
30.3
n/an/actx7 benchmarks
Jan 2024
39.8+9.5
n/an/actx29 benchmarks
Jan 2024
49.2+9.4
n/an/actx27 benchmarks
Feb 2025
$3.00/M in200Kctx
May 2025
44.5-4.7
$3.00/M in200Kctx32 benchmarks
Sep 2025
44.6+0.1
$3.00/M in1.0Mctx(+800K)31 benchmarks
Sep 2025
$1.50/M in(-1.50)1.0Mctx
Feb 2026
52.5+7.9
$3.00/M in(+1.50)1.0Mctx30 benchmarks
Feb 2026
$1.50/M in(-1.50)1.0Mctx
Jun 2026
66.3+13.8
$2.00/M in(+0.50)1.0Mctx17 benchmarks
Jun 2026
$1.00/M in(-1)1.0Mctx
Sep 2026
89.2+22.9
$2.00/M in(+1)1.0Mctx17 benchmarks
Sep 2026
$1.00/M in(-1)1.0Mctx
See the full Claude Sonnet family →
Similar Models
Alibaba Qwen logo
Qwen3.7 Flash
Alibaba Qwen
44.6$0.03/1M
Anthropic logo
Claude Sonnet 4.5
Anthropic
44.6$3.00/1M
Google DeepMind logo
Gemini 2.0 Flash
Google DeepMind
44.4$0.10/1M
Links
Info
AnthropicPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@Anthropic
BenchGecko API
claude-sonnet-4
Specifications
  • Typemultimodal
  • Context200K tokens (~100 books)
  • ReleasedMay 2025
  • LicenseProprietary
  • StatusActive
  • Cost / Message~$0.021
Available On
Anthropic logoAnthropic$3.00
Categories
codingreasoningmathknowledgeagenticgeneralsafetylanguage
Learn More
context-windowtransformertokens
Share & Export
Tweet
Related Models
Alibaba Qwen logoQwen3.7 Flash
44.6
Anthropic logoClaude Sonnet 4.5
44.6
Google DeepMind logoGemini 2.0 Flash
44.4
xAI logoGrok 3
44.6
Claude Sonnet 4 is a proprietary multimodal AI model by Anthropic, released in May 2025. It has an average benchmark score of 44.5. Context window: 200K tokens.

Related Models

Qwen3.7 Flash · Alibaba QwenClaude Sonnet 4.5 · AnthropicGemini 2.0 Flash · Google DeepMindGrok 3 · xAIGemini 2.0 Flash (Dec 2024) · Google DeepMind

Benchmarks

MASKOpenCompass — IFEvalMATH level 5OpenCompass — MMLU-ProLech Mazur Writing

Related Pages

Anthropic · ProviderAnthropic · EconomyAll ModelsCompare ModelsPricingDevelopers · APIContext Window · Glossary

Key facts · as of 2026-10-05

  • Claude Sonnet 4 by Anthropic. BenchGecko score 44.5, rank 176 of 312 scored models (normalized average of public benchmark scores).
  • List price $3.00 input · $15.00 output per 1M tokens (as of 2026-10-05).
  • Sold by 2 providers (as of 2026-10-05): Amazon Bedrock $3.00 in / $15.00 out · Amazon Bedrock $3.00 in / $15.00 out. Every provider

How to cite · data as of 2026-10-05

Claude Sonnet 4 · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-05. https://benchgecko.ai/model/claude-sonnet-4

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP