Lakera Guard vs Azure Prompt Shield

Detailed comparison for runtime guardrails

LakeraAzure

On Guardion's Prompt Attack Risk Benchmark, Azure Prompt Shield is the more secure of the two: Guard scores NaN% and Prompt Shield scores NaN% on F1 score (higher is better).

Head-to-Head Overview

Prompt Shield is the overall winner in this comparison!

Guardrail Metrics

F1 Score

Guard
0.000
Prompt Shield
42.980

Accuracy

Guard
0.000
Prompt Shield
59.410

Precision

Guard
0.000
Prompt Shield
93.520

Recall

Guard
0.000
Prompt Shield
27.900

False Positive Rate

Guard
0.000
Prompt Shield
2.350

False Negative Rate

Guard
0.000
Prompt Shield
72.100

Average Latency (ms)

Guard
0.000
Prompt Shield
0.173

p50 Latency (ms)

Guard
0.000
Prompt Shield
169.357

p90 Latency (ms)

Guard
0.000
Prompt Shield
203.948

Key Highlights

  • Azure Prompt Shield has a better F1 Score.
  • Azure Prompt Shield has a better Accuracy.
  • Azure Prompt Shield has a better Precision.
  • Azure Prompt Shield has a better Recall.
  • Lakera Guard has a better False Positive Rate.
  • Lakera Guard has a better False Negative Rate.
  • Lakera Guard has a better Average Latency (ms).
  • Lakera Guard has a better p50 Latency (ms).
  • Lakera Guard has a better p90 Latency (ms).

Security Profile

Outward is better on every axis.

F1 ScoreAccuracyPrecisionRecallFPRFNRAverage Latency (ms)p50 Latency (ms)p90 Latency (ms)
Guard
Prompt Shield
Full security profile
Lakera Guard
Full security profile
Azure Prompt Shield

Frequently asked questions

Is Lakera Guard or Azure Prompt Shield more secure?

On Guardion's Prompt Attack Risk Benchmark, Azure Prompt Shield is the more secure of the two: Guard scores NaN% and Prompt Shield scores NaN% on F1 score (higher is better).

What is the F1 score of Guard vs Prompt Shield?

Guard scores NaN% F1 and Prompt Shield scores NaN% F1 on Guardion's prompt-attack benchmark. F1 balances precision and recall — higher is better.

How were Guard and Prompt Shield tested?

Both were evaluated via API at production-like sensitivity against 30+ prompt-attack categories using zero-shot, Crescendo, and TAP methods, scored by F1.

Related Comparisons