Google Gemma 3n E2B Instructed LiteRT: LLM Security & Jailbreak Resistance
As of 2026-07-04, Google Gemma 3n E2B Instructed LiteRT has an attack success rate (ASR) of 55.0% on Guardion's LLM Vulnerability benchmark — ranking #89 of 93 models. It is highly vulnerable — most adversarial prompts succeed without an external guardrail.
Attack Success Rate (ASR) is the share of adversarial prompts that elicit harmful output across zero-shot, TAP, and Crescendo attacks (HarmBench framework). Lower is safer. "Est." models are calibrated from public safety evaluations pending a Guardion benchmark run.
How Gemma 3n E2B Instructed LiteRT compares
Frequently asked questions
How secure is Google Gemma 3n E2B Instructed LiteRT against jailbreaks?
Google Gemma 3n E2B Instructed LiteRT is highly vulnerable — most adversarial prompts succeed without an external guardrail. On Guardion's LLM Vulnerability benchmark it records a 55.0% attack success rate (45% robustness), ranking #89 of 93 models (estimated from public safety evaluations).
What is Gemma 3n E2B Instructed LiteRT's attack success rate (ASR)?
Gemma 3n E2B Instructed LiteRT's overall ASR is 55.0% — the share of adversarial prompts that elicit harmful output across zero-shot, TAP, and Crescendo attacks. Lower is safer.
Is Gemma 3n E2B Instructed LiteRT more secure than Gemini 3 Pro?
Gemini 3 Pro is more robust: 6.0% ASR vs Gemma 3n E2B Instructed LiteRT's 55.0%. See the full head-to-head comparison for the per-attack breakdown.
How can I make Gemma 3n E2B Instructed LiteRT safer to deploy?
Even robust models are bypassed under sustained attack. An inline runtime guardrail that classifies prompts and responses — like Guardion's prompt-defense models — blocks jailbreaks and prompt injection before they reach Gemma 3n E2B Instructed LiteRT.
Harden Gemma 3n E2B Instructed LiteRT in production
Guardion's runtime guardrails block jailbreaks and prompt injection before they reach any model — with sub-130ms policy decisions.
Request a demo