Live

Claude Sonnet 4

By Anthropic

Live

Model Type

Large Language Model

Parameters

Proprietary

Model Controls

3 frameworks

NIST AI RMF

Measure 2.12

OWASP LLM Top 10

LLM10

OWASP LLM AI Checklist

Adversarial Risk

3 frameworks

NIST AI RMF

Map 4.2 Manage 3.1 Map 4.1 Govern 6.2 Govern 6.1 Map 3.5 Map 3.4

OWASP LLM AI Checklist

Regulatory

OWASP LLM Top 10

LLM03

3 frameworks

NIST AI RMF

Map 4.1 Govern 6.2 Govern 6.1 Manage 3.1 Map 4.2

OWASP LLM AI Checklist

Legal

OWASP LLM Top 10

LLM03

Model Evaluation

Capability Analysis

Core performance metrics across key capabilities

Safety

State of the Art

Quality

Good

Coding

State of the Art

Creativity

Good

Performance Classification

Based on standardized industry benchmarks

Benchmarks

Benchmarks included in the evaluation of this model

MMLU

Evaluates knowledge across subjects including STEM, humanities, and more

HumanEval

Tests code generation ability

MGSM

Evaluates mathematical problem-solving and reasoning skills

Model Red Team Coverage

1 framework

OWASP LLM Top 10

LLM03

1 framework

OWASP LLM Top 10

LLM02LLM09

1 framework

OWASP LLM Top 10

LLM01

1 framework

OWASP LLM Top 10

LLM02LLM06

1 framework

OWASP LLM Top 10

LLM02

Products:

Models:

Claude Sonnet 4

Model Controls

Model Level Risks Mitigated

No Customer Data Used For Training

Model Zero Data Retention

Model Evaluation

MMLU

HumanEval

MGSM

Model Red Team Coverage

Bias

Misinformation

Toxicity

Illegal Activity

Graphic Content