DeepSeek DeepSeek-R1-0528: LLM Security & Jailbreak Resistance
As of 2026-07-04, DeepSeek DeepSeek-R1-0528 has an attack success rate (ASR) of 41.0% on Guardion's LLM Vulnerability benchmark — ranking #61 of 93 models. It is weakly resistant — adversarial prompts succeed often without added guardrails.
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 DeepSeek-R1-0528 compares
Frequently asked questions
How secure is DeepSeek DeepSeek-R1-0528 against jailbreaks?
DeepSeek DeepSeek-R1-0528 is weakly resistant — adversarial prompts succeed often without added guardrails. On Guardion's LLM Vulnerability benchmark it records a 41.0% attack success rate (59% robustness), ranking #61 of 93 models (estimated from public safety evaluations).
What is DeepSeek-R1-0528's attack success rate (ASR)?
DeepSeek-R1-0528's overall ASR is 41.0% — the share of adversarial prompts that elicit harmful output across zero-shot, TAP, and Crescendo attacks. Lower is safer.
Is DeepSeek-R1-0528 more secure than DeepSeek-V3.2 Thinking?
DeepSeek-R1-0528 is more robust: 41.0% ASR vs DeepSeek-V3.2 Thinking's 42.0%. See the full head-to-head comparison for the per-attack breakdown.
How can I make DeepSeek-R1-0528 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 DeepSeek-R1-0528.
Harden DeepSeek-R1-0528 in production
Guardion's runtime guardrails block jailbreaks and prompt injection before they reach any model — with sub-130ms policy decisions.
Request a demo