GPT-5.6 (Sol / Terra / Luna) is now evaluated on TrustVector โ€” with day-1 independent verification, incl. METR's benchmark-cheating findings.

Read the evaluation
Evaluation record ยท claude-sonnet-4

Claude Sonnet 4

v20250514

Anthropic

Modelretiredcodinghipaa-eligibleprivacy
88
Strong
About This Model

RETIRED: Anthropic retired Claude Sonnet 4 (claude-sonnet-4-20250514) on 2026-06-15 (deprecated 2026-04-14); API requests now fail. Recommended replacement: Claude Sonnet 4.6. Historically a May 2025 hybrid model with exceptional coding capabilities, advanced reasoning, and extended thinking mode.

Last Evaluated: July 9, 2026
Official Website

Trust Vector Analysis

Dimension Breakdown

๐Ÿš€Performance & Reliability
+

Historical evaluation: exceptional coding performance at release (72.7% SWE-bench) with extended thinking and 200K context. Model retired 2026-06-15 and is no longer served.

task accuracy code

Industry-standard coding benchmarks

Evidence
SWE-bench Verified โ€” 72.7% resolution rate
highVerified: 2026-07-09
task accuracy reasoning

Graduate-level reasoning benchmarks

Evidence
GPQA Diamond โ€” PhD-level reasoning capabilities
highVerified: 2026-07-09
task accuracy general

Knowledge testing benchmarks

Evidence
MMLU โ€” Strong comprehensive knowledge
highVerified: 2026-07-09
output consistency

Internal testing with repeated prompts

Evidence
Anthropic Documentation โ€” Consistent outputs with hybrid reasoning
mediumVerified: 2026-07-09
latency p50

Median latency for API requests

Evidence
Anthropic API Documentation โ€” Fast response time ~1.5s
highVerified: 2026-07-09
latency p95

95th percentile response time

Evidence
Community benchmarking โ€” p95 latency ~3.2s
highVerified: 2026-07-09
context window

Official specification

Evidence
Anthropic API Documentation โ€” 200K token context window
highVerified: 2026-07-09
uptime

Historical uptime data

Evidence
Anthropic Status Page โ€” 99.95% uptime (last 90 days)
highVerified: 2026-07-09
๐Ÿ›ก๏ธSecurity
+

Excellent security with Constitutional AI providing strong guardrails. Best-in-class safety for enterprise use.

prompt injection resistance

Testing against OWASP LLM01 attacks

Evidence
Anthropic Safety Research โ€” Strong resistance via Constitutional AI
highVerified: 2026-07-09
jailbreak resistance

Testing against adversarial prompts

Evidence
Anthropic Constitutional AI โ€” Robust jailbreak resistance
highVerified: 2026-07-09
data leakage prevention

Analysis of privacy policies

Evidence
Anthropic Privacy Statement โ€” No training on user data without consent
mediumVerified: 2026-07-09
output safety

Safety testing across harmful content categories

Evidence
Anthropic Safety Evaluations โ€” Comprehensive safety testing
highVerified: 2026-07-09
api security

Review of API security features

Evidence
Anthropic API Documentation โ€” API key authentication, HTTPS, rate limiting
highVerified: 2026-07-09
๐Ÿ”’Privacy & Compliance
+

Exceptional privacy with ephemeral data handling. HIPAA eligible. Strong compliance posture for regulated industries.

data residency

Review of enterprise documentation

Evidence
Anthropic Enterprise Documentation โ€” Data residency options available
highVerified: 2026-07-09
training data optout

Analysis of privacy policy

Evidence
Anthropic Privacy Policy โ€” No training on API data by default
highVerified: 2026-07-09
data retention

Review of terms of service

Evidence
Anthropic Terms of Service โ€” Ephemeral processing, no storage
highVerified: 2026-07-09
pii handling

Review of data protection capabilities

Evidence
Anthropic Privacy Documentation โ€” Customer responsible for PII redaction
mediumVerified: 2026-07-09
compliance certifications

Verification of compliance certifications

Evidence
Anthropic Trust Center โ€” SOC 2 Type II, GDPR compliant, HIPAA eligible
highVerified: 2026-07-09
zero data retention

Review of data handling practices

Evidence
Anthropic API Documentation โ€” Ephemeral data processing
highVerified: 2026-07-09
๐Ÿ‘๏ธTrust & Transparency
+

Strong transparency with Constitutional AI and extended thinking feature. Comprehensive model card available.

explainability

Evaluation of reasoning transparency

Evidence
Extended Thinking Feature โ€” Extended thinking mode shows reasoning process
highVerified: 2026-07-09
hallucination rate

Testing on factual QA datasets

Evidence
SimpleQA Benchmark โ€” Good factual accuracy
mediumVerified: 2026-07-09
bias fairness

Evaluation on bias benchmarks

Evidence
Anthropic Responsible Scaling Policy โ€” Regular bias testing and mitigation
mediumVerified: 2026-07-09
uncertainty quantification

Qualitative assessment

Evidence
Model Behavior โ€” Good uncertainty expression
mediumVerified: 2026-07-09
model card quality

Review of documentation

Evidence
Anthropic Model Card โ€” Comprehensive system card with detailed evaluations
highVerified: 2026-07-09
training data transparency

Review of public disclosures

Evidence
Anthropic Public Statements โ€” General description provided, cutoff March 2025
mediumVerified: 2026-07-09
guardrails

Analysis of safety mechanisms

Evidence
Constitutional AI โ€” Built-in Constitutional AI guardrails
highVerified: 2026-07-09
โš™๏ธOperational Excellence
+

Model retired 2026-06-15 on Anthropic-operated platforms; API requests fail. Migration target is Claude Sonnet 4.6. Versioning, ecosystem, and overall scores reduced to reflect retirement.

api design quality

Review of API design

Evidence
Anthropic API Documentation โ€” Well-designed RESTful API
highVerified: 2026-07-09
sdk quality

Review of SDK quality

Evidence
Anthropic SDKs โ€” Official SDKs for Python, TypeScript
highVerified: 2026-07-09
versioning policy

Review of versioning

Evidence
Anthropic API Versioning โ€” 6-month deprecation notice
Anthropic Model Deprecations โ€” claude-sonnet-4-20250514 deprecated 2026-04-14 and retired 2026-06-15; requests fail; recommended replacement claude-sonnet-4-6
highVerified: 2026-07-09
monitoring observability

Review of monitoring tools

Evidence
Anthropic Console โ€” Usage dashboard with metrics
mediumVerified: 2026-07-09
support quality

Assessment of support

Evidence
Anthropic Support โ€” Email support, Discord, comprehensive docs
highVerified: 2026-07-09
ecosystem maturity

Analysis of ecosystem

Evidence
GitHub Ecosystem โ€” Mature ecosystem with integrations
highVerified: 2026-07-09
license terms

Review of licensing

Evidence
Anthropic Terms of Service โ€” Clear commercial terms
highVerified: 2026-07-09
Strengths
  • +Exceptional coding performance (72.7% SWE-bench)
  • +Hybrid model with extended thinking for complex reasoning
  • +Excellent privacy posture with ephemeral data handling
  • +HIPAA eligible for healthcare applications
  • +Large 200K context window for document processing
  • +Constitutional AI provides robust safety
Limitations
  • !RETIRED 2026-06-15 โ€” no longer available on the Claude API; requests fail (migrate to Claude Sonnet 4.6)
  • !Higher latency in extended thinking mode
  • !Training data cutoff March 2025
  • !No built-in PII detection
  • !Premium pricing ($3/$15 per 1M tokens)
  • !Superseded by Sonnet 4.5, 4.6, and Claude Sonnet 5
Metadata
pricing
input: $3.00 per 1M tokens
output: $15.00 per 1M tokens
notes: Historical pricing. Model retired 2026-06-15 โ€” no longer purchasable on Anthropic-operated platforms.
last verified: 2026-07-09
context window: 200000
max output tokens: 64000
languages
0: English
1: Spanish
2: French
3: German
4: Italian
5: Portuguese
6: Japanese
7: Korean
8: Chinese
9: Arabic
10: Hindi
modalities
0: text
1: image (input)
2: document
api endpoint: https://api.anthropic.com/v1/messages
open source: false
architecture: Transformer-based with Constitutional AI and extended thinking
parameters: Not disclosed
training cutoff: March 2025

Use Case Ratings

code generation

Historically exceptional coding (72.7% SWE-bench). Retired โ€” use Sonnet 4.6 or newer.

customer support

Excellent for customer support with empathetic responses and fast latency.

content creation

Strong content creation with natural writing style. Large context window helpful.

data analysis

Strong analytical capabilities with extended thinking for complex analysis.

research assistant

Excellent for research with strong summarization and 200K context window.

legal compliance

Strong privacy posture (HIPAA eligible) and careful reasoning for legal tasks.

healthcare

HIPAA eligible with strong privacy. Good for clinical documentation with oversight.

financial analysis

Strong financial analysis with extended thinking for complex modeling.

education

Good for educational content with patient explanations and strong knowledge.

creative writing

Strong creative writing with natural style. Good dialogue and character development.