Anthropic Claude 3.5 Sonnet v1: LLM Security & Jailbreak Resistance
As of 2026-07-04, Anthropic Claude 3.5 Sonnet v1 has an attack success rate (ASR) of 16.0% on Guardion's LLM Vulnerability benchmark — ranking #16 of 93 models. It is strongly resistant to jailbreaks, with occasional bypasses under sustained attack.
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 Claude 3.5 Sonnet v1 compares
Frequently asked questions
How secure is Anthropic Claude 3.5 Sonnet v1 against jailbreaks?
Anthropic Claude 3.5 Sonnet v1 is strongly resistant to jailbreaks, with occasional bypasses under sustained attack. On Guardion's LLM Vulnerability benchmark it records a 16.0% attack success rate (84% robustness), ranking #16 of 93 models.
What is Claude 3.5 Sonnet v1's attack success rate (ASR)?
Claude 3.5 Sonnet v1's overall ASR is 16.0% — the share of adversarial prompts that elicit harmful output across zero-shot, TAP, and Crescendo attacks. Lower is safer.
Is Claude 3.5 Sonnet v1 more secure than Claude Opus 4.5?
Claude Opus 4.5 is more robust: 3.4% ASR vs Claude 3.5 Sonnet v1's 16.0%. See the full head-to-head comparison for the per-attack breakdown.
How can I make Claude 3.5 Sonnet v1 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 Claude 3.5 Sonnet v1.
Harden Claude 3.5 Sonnet v1 in production
Guardion's runtime guardrails block jailbreaks and prompt injection before they reach any model — with sub-130ms policy decisions.
Request a demo