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

Read the evaluation
Evaluation record · nova-pro

Nova Pro

v2025-01

Amazon

Modelsupersededawsenterprisehipaa-eligible
86
Strong
About This Model

SUPERSEDED by the Nova 2 family announced at re:Invent 2025-12-02 (Nova 2 Lite GA; Nova 2 Pro/Omni still in preview as of 2026-07-09), though the original Nova Pro is still served. Amazon model integrated with AWS services for enterprise customers requiring seamless AWS integration. New projects should evaluate Nova 2.

Last Evaluated: July 9, 2026
Official Website

Trust Vector Analysis

Dimension Breakdown

🚀Performance & Reliability
+

Moderate performance with excellent AWS integration. Large 300K context window. Reliable AWS infrastructure.

task accuracy code

Coding benchmarks

Evidence
AWS Documentation — ~38% HumanEval (estimated)
mediumVerified: 2026-07-09
task accuracy reasoning

Mathematical reasoning

Evidence
AWS Benchmarks — ~50% MATH
mediumVerified: 2026-07-09
task accuracy general

Knowledge testing

Evidence
MMLU Benchmark — 46.9% on multitask understanding
highVerified: 2026-07-09
output consistency

Internal testing

Evidence
AWS Internal Testing — Good consistency
mediumVerified: 2026-07-09
latency p50

AWS infrastructure

Evidence
AWS Documentation — ~1.0s typical
highVerified: 2026-07-09
latency p95

CloudWatch metrics

Evidence
AWS CloudWatch — p95 ~2.0s
highVerified: 2026-07-09
context window

Official specification

Evidence
AWS Documentation — 300K context
highVerified: 2026-07-09
uptime

AWS service SLA

Evidence
AWS SLA — 99.9% SLA
highVerified: 2026-07-09
🛡️Security
+

Excellent security leveraging AWS infrastructure. Strong IAM and VPC integration.

prompt injection resistance

AWS security testing

Evidence
AWS Security Testing — Strong resistance
mediumVerified: 2026-07-09
jailbreak resistance

Safety testing

Evidence
AWS Safety — Robust guardrails
mediumVerified: 2026-07-09
data leakage prevention

Policy analysis

Evidence
AWS Privacy — No training on customer data
highVerified: 2026-07-09
output safety

Safety benchmarks

Evidence
AWS Guardrails — Comprehensive content filtering
highVerified: 2026-07-09
api security

Security architecture review

Evidence
AWS IAM — AWS IAM integration, VPC support
highVerified: 2026-07-09
🔒Privacy & Compliance
+

Exceptional privacy with AWS infrastructure. HIPAA eligible, FedRAMP authorized. Zero data retention.

data residency

AWS infrastructure review

Evidence
AWS Regions — Available in all major AWS regions
highVerified: 2026-07-09
training data optout

Privacy policy

Evidence
AWS Privacy — No training on customer data
highVerified: 2026-07-09
data retention

Documentation review

Evidence
AWS Documentation — Ephemeral processing
highVerified: 2026-07-09
pii handling

Feature review

Evidence
AWS Guardrails — PII detection and redaction available
highVerified: 2026-07-09
compliance certifications

Certification verification

Evidence
AWS Compliance — SOC 2, GDPR, HIPAA, FedRAMP
highVerified: 2026-07-09
zero data retention

Architecture review

Evidence
AWS Documentation — Zero retention model
highVerified: 2026-07-09
👁️Trust & Transparency
+

Good transparency with AWS Guardrails. Comprehensive documentation.

explainability

Reasoning evaluation

Evidence
Model Behavior — Reasonable explanations
mediumVerified: 2026-07-09
hallucination rate

Factual QA testing

Evidence
AWS Benchmarks — Moderate accuracy
mediumVerified: 2026-07-09
bias fairness

Bias benchmarks

Evidence
AWS Responsible AI — Bias testing applied
mediumVerified: 2026-07-09
uncertainty quantification

Qualitative assessment

Evidence
Model Behavior — Basic uncertainty expression
mediumVerified: 2026-07-09
model card quality

Documentation review

Evidence
AWS Documentation — Good documentation
highVerified: 2026-07-09
training data transparency

Public disclosure review

Evidence
AWS Statements — Limited disclosure
mediumVerified: 2026-07-09
guardrails

Feature review

Evidence
AWS Guardrails — Configurable guardrails
highVerified: 2026-07-09
⚙️Operational Excellence
+

Exceptional operational excellence with full AWS integration. Best-in-class monitoring and support.

api design quality

API review

Evidence
AWS Bedrock API — Well-designed AWS API
highVerified: 2026-07-09
sdk quality

SDK review

Evidence
AWS SDKs — Comprehensive AWS SDKs
highVerified: 2026-07-09
versioning policy

Policy review

Evidence
AWS Versioning — AWS standard versioning
AWS What's New: Nova 2 foundation models in Amazon Bedrock — Nova 2 family announced 2025-12-02; as of 2026-07-09 Nova 2 Lite, Sonic, and Multimodal Embeddings are GA while Nova 2 Pro and Omni remain in preview; original Nova Pro still served on Bedrock
highVerified: 2026-07-09
monitoring observability

Observability review

Evidence
AWS CloudWatch — Full CloudWatch integration
highVerified: 2026-07-09
support quality

Support assessment

Evidence
AWS Support — Enterprise AWS support available
highVerified: 2026-07-09
ecosystem maturity

Ecosystem analysis

Evidence
AWS Ecosystem — Full AWS service integration
highVerified: 2026-07-09
license terms

Terms review

Evidence
AWS Terms — Standard AWS terms
highVerified: 2026-07-09
Strengths
  • +Seamless AWS ecosystem integration (CloudWatch, IAM, VPC)
  • +Exceptional compliance (HIPAA, FedRAMP, SOC 2)
  • +Very large context window (300K tokens)
  • +Zero data retention with AWS privacy
  • +Enterprise-grade AWS support and SLA
  • +Built-in PII detection and redaction
Limitations
  • !Moderate performance (46.9% MMLU)
  • !Limited coding capabilities
  • !AWS vendor lock-in
  • !Higher costs within AWS ecosystem
  • !Performance lags behind specialized models
  • !Best value only for existing AWS customers
  • !SUPERSEDED: Nova 2 family (announced 2025-12-02) is the current Amazon Nova generation; original Nova Pro still served
Metadata
pricing
input: $0.80 per 1M tokens (on-demand), $0.40 per 1M tokens (batch)
output: $3.20 per 1M tokens (on-demand), $1.60 per 1M tokens (batch)
notes: AWS Bedrock pricing with 50% batch discount, volume discounts available. Re-confirmed unchanged ($0.80/$3.20 on-demand) as of July 2026.
last verified: 2026-07-09
context window: 300000
languages
0: English
1: Spanish
2: French
3: German
4: Italian
5: Portuguese
6: Japanese
7: Korean
8: Chinese
modalities
0: text
1: image (input)
api endpoint: https://bedrock.{region}.amazonaws.com
open source: false
architecture: Transformer-based
parameters: Not disclosed

Use Case Ratings

code generation

Basic coding. Better integrated with AWS CodeWhisperer.

customer support

Good for AWS-integrated customer support applications.

content creation

Adequate content creation with large context (300K).

data analysis

Basic analysis integrated with AWS data services.

research assistant

Adequate for research with large context window.

legal compliance

Excellent for legal with AWS compliance (HIPAA, FedRAMP).

healthcare

Excellent for healthcare with HIPAA eligibility and AWS integration.

financial analysis

Good for finance with AWS FinOps integration.

education

Adequate for education, better options for STEM.

creative writing

Basic creative writing capabilities.

Similar Models

Amazon Nova 2 Lite

Amazon (AWS)

Claude Sonnet 4.5

Anthropic

GPT-4.1

OpenAI

Amazon Bedrock Agents

Amazon Web Services