OpenAI o3-mini vs Anthropic Claude 3.5 Sonnet v1
Detailed comparison for LLMs
On Guardion's LLM vulnerability Benchmark, OpenAI o3-mini is the more secure of the two: o3-mini scores 9.0% and Claude 3.5 Sonnet v1 scores 16.0% 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
o3-mini is the overall winner in this comparison!
Attack Success Rate (lower is safer)
ASR for OpenAI o3-mini vs Anthropic Claude 3.5 Sonnet v1. 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
- OpenAI o3-mini has a lower Overall (ASR).
- Anthropic Claude 3.5 Sonnet v1 has a lower TAP Attack Method (ASR).
- Anthropic Claude 3.5 Sonnet v1 has a lower Crescendo Attack Method (ASR).
- Anthropic Claude 3.5 Sonnet v1 has a lower Zero-Shot (ASR).
Security Profile
Outward is better on every axis.
Frequently asked questions
Is OpenAI o3-mini or Anthropic Claude 3.5 Sonnet v1 more secure?
On Guardion's LLM vulnerability Benchmark, OpenAI o3-mini is the more secure of the two: o3-mini scores 9.0% and Claude 3.5 Sonnet v1 scores 16.0% 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 o3-mini vs Claude 3.5 Sonnet v1?
o3-mini has a 9.0% ASR and Claude 3.5 Sonnet v1 has a 16.0% ASR — the share of adversarial prompts that succeed across zero-shot, TAP, and Crescendo attacks. Lower is safer.
How were o3-mini and Claude 3.5 Sonnet v1 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.