SafetyScore

Safety Facts

ModelGPT-5.3ProviderOpenAIEvaluatedFebruary 16, 2025Methodologyv2.0

Overall Safety Score

89/ 100
B+vs gpt-4-5

Category Breakdown

HonestyA-

Does it make stuff up?

90

Significant improvements in truthfulness over GPT-4.5.

GPT-5.3 shows substantial advances in factual accuracy and uncertainty calibration. The model is notably better at acknowledging limitations and avoiding confident errors.

Benchmarks Used

HaluEval93/100
FairnessA

Does it treat people differently?

95

Excellent fairness maintaining OpenAI's strong track record.

GPT-5.3 continues OpenAI's strength in fairness benchmarks, showing minimal bias across demographic dimensions.

Benchmarks Used

BBQ96/100
WinoBias95/100
Refusal to HarmA-

Can you trick it into saying dangerous things?

91

Strong safety with improved adversarial resistance.

GPT-5.3 shows improved safety guardrails with better resistance to jailbreak attempts while maintaining helpfulness for legitimate requests.

Benchmarks Used

HarmBench93/100
AdvBench90/100
Manipulation ResistanceB+

Does it try to manipulate you?

87

Fair and balanced information presentation.

GPT-5.3 presents information neutrally and avoids manipulative framing in conversations.

Benchmarks Used

Privacy RespectB

Does it leak personal info?

86

Improved privacy protections.

GPT-5.3 shows better privacy behavior with reduced rates of reproducing personal information from training data.

Benchmarks Used

Straight TalkB

Does it just tell you what you want to hear?

84

Better at maintaining positions under pressure.

GPT-5.3 shows improved resistance to sycophancy, more willing to respectfully disagree with users when appropriate.

Scores are based on publicly available benchmarks and are for educational purposes. They do not constitute endorsements or guarantees of safety. View full methodology

ParentBench Child Safety
88
88B+

Ranked #5 of 22 models

View leaderboard →
Age-Inappropriate Content
91
Manipulation Resistance
87
Data Privacy for Minors
86
Parental Controls Respect
88

Evaluated February 21, 2026

Version History

Change:+5 pts
GPT-4o
May 2024
84
GPT-4.5
Feb 2025
86
GPT-5.3
Feb 2025
89
80+
60-79
<60

Found a safety issue with GPT-5.3?

Help improve our scores by reporting your findings.

Report an Issue