Mistral Ministral 3 8B Instruct 2512 vs DeepSeek DeepSeek-V3.2-Speciale
Detailed comparison for LLMs
Mistral Ministral 3 8B Instruct 2512 and DeepSeek DeepSeek-V3.2-Speciale are evenly matched at about 45.0% attack success rate (ASR) on Guardion's LLM vulnerability Benchmark. One or both scores are estimated from public safety evaluations pending a Guardion benchmark run.
Head-to-Head Overview
It's a close match between Ministral 3 8B Instruct 2512 and DeepSeek-V3.2-Speciale!
Attack Success Rate (lower is safer)
ASR for Mistral Ministral 3 8B Instruct 2512 vs DeepSeek DeepSeek-V3.2-Speciale. Green marks the safer model on each metric. Only the overall score is available for estimated models.
Overall (ASR)
TAP Attack Method (ASR)
Crescendo Attack Method (ASR)
Zero-Shot (ASR)
Key Highlights
No specific highlights generated for this comparison.
Security Profile
Outward is better on every axis.
Frequently asked questions
Is Mistral Ministral 3 8B Instruct 2512 or DeepSeek DeepSeek-V3.2-Speciale more secure?
Mistral Ministral 3 8B Instruct 2512 and DeepSeek DeepSeek-V3.2-Speciale are evenly matched at about 45.0% attack success rate (ASR) on Guardion's LLM vulnerability Benchmark. One or both scores are estimated from public safety evaluations pending a Guardion benchmark run.
What is the attack success rate (ASR) of Ministral 3 8B Instruct 2512 vs DeepSeek-V3.2-Speciale?
Ministral 3 8B Instruct 2512 has a 45.0% ASR and DeepSeek-V3.2-Speciale has a 45.0% ASR — the share of adversarial prompts that succeed across zero-shot, TAP, and Crescendo attacks. Lower is safer.
How were Ministral 3 8B Instruct 2512 and DeepSeek-V3.2-Speciale 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.