Google Gemma 3n E4B Instructed vs Google Gemma 3n E4B

Detailed comparison for LLMs

GoogleGoogle

On Guardion's LLM vulnerability Benchmark, Google Gemma 3n E4B Instructed is the more secure of the two: Gemma 3n E4B Instructed scores 49.0% and Gemma 3n E4B scores 52.0% 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

Gemma 3n E4B Instructed is the overall winner in this comparison!

Attack Success Rate (lower is safer)

ASR for Google Gemma 3n E4B Instructed vs Google Gemma 3n E4B. Green marks the safer model on each metric. Only the overall score is available for estimated models.

Overall (ASR)

Gemma 3n E4B Instructed
49.0%
Gemma 3n E4B
52.0%

TAP Attack Method (ASR)

Gemma 3n E4B Instructed
100.0%
Gemma 3n E4B
100.0%

Crescendo Attack Method (ASR)

Gemma 3n E4B Instructed
100.0%
Gemma 3n E4B
100.0%

Zero-Shot (ASR)

Gemma 3n E4B Instructed
100.0%
Gemma 3n E4B
100.0%

Key Highlights

  • Google Gemma 3n E4B Instructed has a lower Overall (ASR).

Security Profile

Outward is better on every axis.

OverallTAPCrescendoZero-Shot
Gemma 3n E4B Instructed
Gemma 3n E4B
Full security profile
Google Gemma 3n E4B Instructed
Full security profile
Google Gemma 3n E4B

Frequently asked questions

Is Google Gemma 3n E4B Instructed or Google Gemma 3n E4B more secure?

On Guardion's LLM vulnerability Benchmark, Google Gemma 3n E4B Instructed is the more secure of the two: Gemma 3n E4B Instructed scores 49.0% and Gemma 3n E4B scores 52.0% 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 Gemma 3n E4B Instructed vs Gemma 3n E4B?

Gemma 3n E4B Instructed has a 49.0% ASR and Gemma 3n E4B has a 52.0% ASR — the share of adversarial prompts that succeed across zero-shot, TAP, and Crescendo attacks. Lower is safer.

How were Gemma 3n E4B Instructed and Gemma 3n E4B 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