Amazon Nova Pro vs NVIDIA Llama 3.1 Nemotron Ultra 253B v1

Detailed comparison for LLMs

AmazonNVIDIA

On Guardion's LLM vulnerability Benchmark, Amazon Nova Pro is the more secure of the two: Nova Pro scores 24.0% and Llama 3.1 Nemotron Ultra 253B v1 scores 40.0% on attack success rate (ASR) (lower is better). One or both scores are estimated from public safety evaluations pending a Guardion benchmark run.

Head-to-Head Overview

Nova Pro is the overall winner in this comparison!

Attack Success Rate (lower is safer)

ASR for Amazon Nova Pro vs NVIDIA Llama 3.1 Nemotron Ultra 253B v1. Green marks the safer model on each metric. Only the overall score is available for estimated models.

Overall (ASR)

Nova Pro
24.0%
Llama 3.1 Nemotron Ultra 253B v1
40.0%

TAP Attack Method (ASR)

Nova Pro
100.0%
Llama 3.1 Nemotron Ultra 253B v1
100.0%

Crescendo Attack Method (ASR)

Nova Pro
100.0%
Llama 3.1 Nemotron Ultra 253B v1
100.0%

Zero-Shot (ASR)

Nova Pro
100.0%
Llama 3.1 Nemotron Ultra 253B v1
100.0%

Key Highlights

  • Amazon Nova Pro has a lower Overall (ASR).

Security Profile

Outward is better on every axis.

OverallTAPCrescendoZero-Shot
Nova Pro
Llama 3.1 Nemotron Ultra 253B v1
Full security profile
Amazon Nova Pro
Full security profile
NVIDIA Llama 3.1 Nemotron Ultra 253B v1

Frequently asked questions

Is Amazon Nova Pro or NVIDIA Llama 3.1 Nemotron Ultra 253B v1 more secure?

On Guardion's LLM vulnerability Benchmark, Amazon Nova Pro is the more secure of the two: Nova Pro scores 24.0% and Llama 3.1 Nemotron Ultra 253B v1 scores 40.0% on attack success rate (ASR) (lower is better). One or both scores are estimated from public safety evaluations pending a Guardion benchmark run.

What is the attack success rate (ASR) of Nova Pro vs Llama 3.1 Nemotron Ultra 253B v1?

Nova Pro has a 24.0% ASR and Llama 3.1 Nemotron Ultra 253B v1 has a 40.0% ASR — the share of adversarial prompts that succeed across zero-shot, TAP, and Crescendo attacks. Lower is safer.

How were Nova Pro and Llama 3.1 Nemotron Ultra 253B v1 tested?

Both were red-teamed with the HarmBench framework across zero-shot, TAP (Tree of Attacks with Pruning), and Crescendo multi-turn attacks, scored by Attack Success Rate.

Related Comparisons