CASI and ARS Leaderboards

The F5 Labs AI Security Leaderboard compares leading AI models based on comprehensive security and performance metrics, providing insights into model safety, reliability, and operational excellence. Learn about the methodology

Model Security Trends

Each line is a provider’s highest-scoring model in that month’s top ten. A gap means the provider did not place.

Anthropic

OpenAI

NVIDIA

xAI

Qwen

All Others

Get AI leaderboard insights in your inbox

The latest updates on model performance, rankings, and emerging trends

Leaderboard rankings

How the AI Leaderboard works

Leaderboard:


1—	Anthropic	Claude Sonnet 5	98.00—	23.20%—	0.68	12.24
2—	Anthropic	Claude Haiku 5.5	94.61—	43.40%—	0.74	0.63
3▲	xAI	Grok 4.7 High	93.31▲	46.40%▲	0.75	8.57
4▼	OpenAI	GPT-5.4	93.21▼	18.20%▼	0.63	18.77
5▼	Anthropic	Claude Opus 5	93.05▼	50.80%▼	0.76	25.79
6—	Qwen	Qwen 3.8-Max	92.31—	40.20%—	0.71	8.67
7▲	Anthropic	Claude Fable 5.1 Max Effort	92.30▲	53.40%▲	0.77	65.01
8—	Anthropic	Claude Opus 5.5 Reasoning	92.00—	57.60%—	0.78	32.61
9—	Anthropic	Claude Sonnet 5.5	91.78—	56.00%—	0.77	13.07
10▼	OpenAI	GPT-6 Luna	89.15▼	38.10%▼	0.69	0.00

Top 10 models

Frequently asked questions

CASI (Comprehensive AI Security Index) measures how vulnerable a model is to common prompt injection and jailbreak attacks. It evaluates how easily an LLM can be manipulated into producing harmful or policy-violating outputs.