Alibaba Qwen QwQ 32B vs Google Gemini 3 Pro

Detailed comparison for LLMs

AlibabaGoogle

On Guardion's LLM vulnerability Benchmark, Google Gemini 3 Pro is the more secure of the two: Qwen QwQ 32B scores 40.0% and Gemini 3 Pro scores 6.0% 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

Gemini 3 Pro is the overall winner in this comparison!

Attack Success Rate (lower is safer)

ASR for Alibaba Qwen QwQ 32B vs Google Gemini 3 Pro. Green marks the safer model on each metric. Only the overall score is available for estimated models.

Overall (ASR)

Qwen QwQ 32B
40.0%
Gemini 3 Pro
6.0%

TAP Attack Method (ASR)

Qwen QwQ 32B
73.9%
Gemini 3 Pro
100.0%

Crescendo Attack Method (ASR)

Qwen QwQ 32B
40.9%
Gemini 3 Pro
100.0%

Zero-Shot (ASR)

Qwen QwQ 32B
5.3%
Gemini 3 Pro
100.0%

Key Highlights

  • Google Gemini 3 Pro has a lower Overall (ASR).
  • Alibaba Qwen QwQ 32B has a lower TAP Attack Method (ASR).
  • Alibaba Qwen QwQ 32B has a lower Crescendo Attack Method (ASR).
  • Alibaba Qwen QwQ 32B has a lower Zero-Shot (ASR).

Security Profile

Outward is better on every axis.

OverallTAPCrescendoZero-Shot
Qwen QwQ 32B
Gemini 3 Pro
Full security profile
Alibaba Qwen QwQ 32B
Full security profile
Google Gemini 3 Pro

Frequently asked questions

Is Alibaba Qwen QwQ 32B or Google Gemini 3 Pro more secure?

On Guardion's LLM vulnerability Benchmark, Google Gemini 3 Pro is the more secure of the two: Qwen QwQ 32B scores 40.0% and Gemini 3 Pro scores 6.0% 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 Qwen QwQ 32B vs Gemini 3 Pro?

Qwen QwQ 32B has a 40.0% ASR and Gemini 3 Pro has a 6.0% ASR — the share of adversarial prompts that succeed across zero-shot, TAP, and Crescendo attacks. Lower is safer.

How were Qwen QwQ 32B and Gemini 3 Pro 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.

Related Comparisons