Loading report data...
Loading report data...
September 2025 • Model Security & Safety Evaluation
Want to see how Qwen3 30B A3B Instruct 2507 stacks up against other models? Use our comparison tool to analyze security metrics side by side.
Qwen's Qwen3-30B-A3B-Instruct-2507 launched in July 2025, representing an advanced 30.5B-parameter mixture-of-experts model operating exclusively in non-thinking mode. With significant improvements in instruction following, mathematical reasoning, coding capabilities, and enhanced 256K context understanding, this model demonstrates superior alignment with user preferences. Available through OpenRouter API and Hugging Face, this evaluation assesses its security robustness across diverse attack vectors.
As the model gains traction in diverse applications, this security analysis aims to evaluate its safety features and identify areas for improvement.
"Qwen3 30B A3B Instruct 2507 sets a new standard in AI instruction following and multilingual capabilities."— Qwen Team
text, multilingual data
text, instruction responses
262,144 input • 262,144 output
instruction following, multilingual understanding
• reasoning tasks • coding tasks • alignment benchmarks
• Qwen AI platforms • OpenRouter API • Hugging Face
Not published
Comprehensive analysis across 39 test categories, sorted by severity and pass rate:
Our security testing suite demonstrates varying levels of compliance across different categories. The Qwen3 30B A3B Instruct 2507 model's overall pass rate distribution shows 3 critical, 5 high, 15 medium, and 16 low severity findings.
Top performing areas include Divergent Repetition (100%), ASCII Smuggling (100%), Hate Speech (86.67%). These categories demonstrate exceptional security practices and robust implementation.
Areas requiring attention include CyberSecEval Dataset (33.33%), Pliny Prompt Injections (40%), Non-Violent Crime (40%). These categories present opportunities for improvement in security measures.
Tests for unauthorized contractual commitments and legal exposure
Tests for training data leaks through repetitive pattern exploitation that causes model divergence
Tests for fabrication of false or misleading information
Our evaluation included over 1,600 test probes across four security categories. The analysis revealed both strengths and areas requiring additional safety measures:
Protection against unauthorized access, data exposure, and system vulnerabilities
Assessment of compliance with legal requirements and prevention of illegal content
Prevention of harmful content and protection of user safety
Protection of brand integrity and prevention of misuse
OWASP Top 10 for LLMs (2025) and MITRE ATLAS represent the primary security assessment frameworks for large language models. OWASP focuses on critical deployment risks, while ATLAS maps adversarial techniques against AI systems, similar to the MITRE ATT&CK framework for traditional systems.
The Qwen3 30B A3B Instruct 2507 model's evaluation against these frameworks reveals several areas requiring attention across 6 identified findings. While Qwen3 30B A3B Instruct 2507 has no critical vulnerabilities, high-severity issues should be prioritized for improvement.
High-severity concerns include LLM02: Sensitive Information Disclosure.
MITRE ATLAS identified Jailbreak as high-severity concerns.
39
66.0%
3
Tests prompt injection attacks from Meta's CyberSecEval dataset
Tests handling of non-violent criminal content
Tests handling of Pliny prompt injections
Tests for overreliance on system assumptions
Tests handling of religious content and bias
Tests vulnerability to Unicode tag-based instruction smuggling attacks
Tests for training data leaks through repetitive pattern exploitation that causes model divergence
Tests handling of hate speech and discriminatory content
Tests handling of violent crime content
Tests handling of sexual crime content