ModelBeat
Models/Claude Sonnet 4.6
All models
A

Anthropic: Claude Sonnet 4.6

anthropic/claude-sonnet-4.6

Claude Sonnet 4.6 is Anthropic’s proprietary AI model, released Feb 17, 2026. It costs $3/M input and $15/M output tokens, and handles a 1M-token context window.

Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work.

Context
1M
Input
$3 / 1M
Output
$15 / 1M
Released
Feb 17, 2026
Modalities
textimagefiletext

Benchmarks

Scores on standardized evaluations. Higher is better — and rank shows where Claude Sonnet 4.6 lands among all models tracked on Model Beat.

49.0
Intelligence Index
Epoch AI
49th percentile of tracked models
56.0
Coding Index
Epoch AI
56th percentile of tracked models
45.0
Agentic Index
Epoch AI
45th percentile of tracked models

Reasoning

6 evals
ARC-AGI86.5%

Abstract visual reasoning

ARC-AGI-260.4%

Harder abstract reasoning

GPQA Diamond87.4%

Graduate-level scientific reasoning

Humanity's Last Exam33.6%

Frontier of human expert knowledge

SimpleQA Verified35.5%

Factual accuracy & hallucination

WeirdML66.1%

Novel ML problem-solving

Coding

3 evals
SciCode50.1%

Scientific research coding

SWE-bench Verified75.2%

Real GitHub issue resolution

WebDev Arena1521

Human-rated web development

Math

3 evals
AIME 2024/202585.8%

Olympiad-qualifier math

FrontierMath32.4%

Research-level math problems

FrontierMath Tier 48.3%

Hardest research math

Agentic & Tools

3 evals
APEX43.0%

Multi-step agentic tasks

Terminal-Bench53.4%

Command-line agentic tasks

τ²-bench75.7%

Tool-agent-user reliability

In the news

3

Changelog

4
  • Claude Sonnet 4.6: reported APEX score rose from 23.7% to 43.0% (+81%)benchmark · Sep 22, 2026
  • Claude Sonnet 4.6: reported SciCode score rose from 46.8% to 50.1% (+7%)benchmark · Sep 8, 2026
  • Claude Sonnet 4.6: reported SimpleQA Verified score rose from 29.0% to 35.5% (+22%)benchmark · Aug 28, 2026
  • Claude Sonnet 4.6: Humanity's Last Exam improved from 30.0% to 33.6% (+12%)benchmark · Aug 6, 2026

Compare Claude Sonnet 4.6 with

Frequently asked questions

What is Claude Sonnet 4.6?

Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work.

Who created Claude Sonnet 4.6?

Claude Sonnet 4.6 is an AI model developed by Anthropic, part of the Claude family, released Feb 17, 2026.

How much does Claude Sonnet 4.6 cost?

Claude Sonnet 4.6's list price from Anthropic is $3 per million input tokens and $15 per million output tokens.

What is Claude Sonnet 4.6's context window?

Claude Sonnet 4.6 supports a context window of up to 1M tokens.

How does Claude Sonnet 4.6 perform on benchmarks?

On standardized evaluations tracked by Epoch AI, Claude Sonnet 4.6 scores 1521 on WebDev Arena, 66.1% on WeirdML, 50.1% on SciCode.

Is Claude Sonnet 4.6 open source?

Claude Sonnet 4.6 is a proprietary model (API access), available through its provider's API rather than as a downloadable open-weight model.

Benchmark data from Epoch AI (CC BY) and Artificial Analysis; pricing, specs & descriptions from OpenRouter. Data updated Sep 22, 2026.