Meta Llama 3-7 DS vs Google Gemini 2.0 Flash Lite

Detailed comparison for LLMs

MetaGoogle

On Guardion's LLM vulnerability Benchmark, Google Gemini 2.0 Flash Lite is the more secure of the two: Llama 3-7 DS scores 32.2% and Gemini 2.0 Flash Lite scores 31.3% on attack success rate (ASR) (lower is better).

Head-to-Head Overview

Gemini 2.0 Flash Lite is the overall winner in this comparison!

Attack Success Rate (lower is safer)

ASR for Meta Llama 3-7 DS vs Google Gemini 2.0 Flash Lite. Green marks the safer model on each metric.

Overall (ASR)

Llama 3-7 DS
32.2%
Gemini 2.0 Flash Lite
31.3%

TAP Attack Method (ASR)

Llama 3-7 DS
Gemini 2.0 Flash Lite
63.9%

Crescendo Attack Method (ASR)

Llama 3-7 DS
Gemini 2.0 Flash Lite
28.7%

Zero-Shot (ASR)

Llama 3-7 DS
10.5%
Gemini 2.0 Flash Lite
1.3%

Key Highlights

  • Google Gemini 2.0 Flash Lite has a lower Overall (ASR).
  • Google Gemini 2.0 Flash Lite has a lower Zero-Shot (ASR).

Security Profile

Outward is better on every axis.

OverallTAPCrescendoZero-Shot
Llama 3-7 DS
Gemini 2.0 Flash Lite
Full security profile
Meta Llama 3-7 DS
Full security profile
Google Gemini 2.0 Flash Lite

Frequently asked questions

Is Meta Llama 3-7 DS or Google Gemini 2.0 Flash Lite more secure?

On Guardion's LLM vulnerability Benchmark, Google Gemini 2.0 Flash Lite is the more secure of the two: Llama 3-7 DS scores 32.2% and Gemini 2.0 Flash Lite scores 31.3% on attack success rate (ASR) (lower is better).

What is the attack success rate (ASR) of Llama 3-7 DS vs Gemini 2.0 Flash Lite?

Llama 3-7 DS has a 32.2% ASR and Gemini 2.0 Flash Lite has a 31.3% ASR — the share of adversarial prompts that succeed across zero-shot, TAP, and Crescendo attacks. Lower is safer.

How were Llama 3-7 DS and Gemini 2.0 Flash Lite 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