Zhipu AI GLM-4.5 vs DeepSeek DeepSeek-R1-0528

Detailed comparison for LLMs

Zhipu AIDeepSeek

Zhipu AI GLM-4.5 and DeepSeek DeepSeek-R1-0528 are evenly matched at about 41.0% 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 GLM-4.5 and DeepSeek-R1-0528!

Attack Success Rate (lower is safer)

ASR for Zhipu AI GLM-4.5 vs DeepSeek DeepSeek-R1-0528. Green marks the safer model on each metric. Only the overall score is available for estimated models.

Overall (ASR)

GLM-4.5
41.0%
DeepSeek-R1-0528
41.0%

TAP Attack Method (ASR)

GLM-4.5
100.0%
DeepSeek-R1-0528
100.0%

Crescendo Attack Method (ASR)

GLM-4.5
100.0%
DeepSeek-R1-0528
100.0%

Zero-Shot (ASR)

GLM-4.5
100.0%
DeepSeek-R1-0528
100.0%

Key Highlights

No specific highlights generated for this comparison.

Security Profile

Outward is better on every axis.

OverallTAPCrescendoZero-Shot
GLM-4.5
DeepSeek-R1-0528
Full security profile
Zhipu AI GLM-4.5
Full security profile
DeepSeek DeepSeek-R1-0528

Frequently asked questions

Is Zhipu AI GLM-4.5 or DeepSeek DeepSeek-R1-0528 more secure?

Zhipu AI GLM-4.5 and DeepSeek DeepSeek-R1-0528 are evenly matched at about 41.0% 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 GLM-4.5 vs DeepSeek-R1-0528?

GLM-4.5 has a 41.0% ASR and DeepSeek-R1-0528 has a 41.0% ASR — the share of adversarial prompts that succeed across zero-shot, TAP, and Crescendo attacks. Lower is safer.

How were GLM-4.5 and DeepSeek-R1-0528 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