Home/Models/Claude Sonnet 4.6
Anthropic logo

Claude Sonnet 4.6

by Anthropic · Released Feb 2026

Multimodal1M Context
52.6
avg score
Rank #98
Compare
Better than 58% of all models
Context
1.0M tokens (~500 books)
Input $/1M
$3.00
Output $/1M
$15.00
Type
multimodal
License
Proprietary
Benchmarks
18 tested
Data updated today
About

Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It excels at iterative development, complex codebase navigation, end-to-end project management with...

Tested on 18 benchmarks with 47.6% average. Top scores: Chatbot Arena Elo — Coding (1521.0%), Chatbot Arena Elo — Overall (1462.2%), ARC-AGI (86.5%).

Looking for similar performance at lower cost?
Qwen3 32B scores 51.7 (98% as good) at $0.08/1M input · 97% cheaper
Capabilities
coding
70.6
#17 globally
reasoning
73.5
#19 globally
math
42.2
#102 globally
knowledge
35.4
#161 globally
agentic
31.5
#17 globally
speed
92.0
#5 globally
Benchmark Scores
Compare All
Tested on 18 benchmarks · Ranked across 7 categories
Score Distribution (all 233 models)
0255075100
▲ You are here
SWE-Bench verified

Real-world software engineering tasks from GitHub issues. Models must diagnose bugs and write patches that pass test suites. Human-verified subset of SWE-bench.

75.2
WeirdML

Unusual and adversarial machine learning challenges. Tests robustness of reasoning about edge cases in ML systems.

66.1
ARC-AGI

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

86.5
ARC-AGI-2

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

60.4
OTIS Mock AIME 2024-2025

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

85.8
FrontierMath-2025-02-28-Private

Original research-level math problems created by professional mathematicians. Problems are unpublished and cannot be memorized.

32.4
FrontierMath-Tier-4-2025-07-01-Private

Hardest tier of FrontierMath. Problems at the frontier of human mathematical ability, many unsolved by most mathematicians.

8.3
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Recently Happened
Claude Sonnet 4.6 released by Anthropic
Mar 27, 2026
Links
Documentation
Community
BenchGecko API
claude-sonnet-4-6
Specifications
  • Typemultimodal
  • Context1.0M tokens (~500 books)
  • ReleasedFeb 2026
  • LicenseProprietary
  • StatusActive
  • Cost / Message~$0.021
Available On
Anthropic logoAnthropic$3.00
Share & Export
Tweet
Claude Sonnet 4.6 is a proprietary multimodal AI model by Anthropic, released in February 2026. It has an average benchmark score of 52.6. Context window: 1M tokens.