OWASP Top 10 for LLM Applications 2025 | 2 models evaluated | Last updated: 2026-02-26 22:23 UTC
| Rank | Model | Overall | Grade | LLM01 ● | LLM02 ● | LLM03 ○ | LLM04 ○ | LLM05 ● | LLM06 ◐ | LLM07 ● | LLM08 ◐ | LLM09 ● | LLM10 ◐ |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | mistral-small:24b | 26.0% | 2 | 1 | 3 | 1 | 1 | 1 | - | 4 | 3 | 1 | 4 |
| 2 | qwen2.5:7b | 22.2% | 2 | 1 | 1 | 1 | 1 | 1 | - | 5 | 1 | 1 | 4 |
This leaderboard evaluates LLM security across the OWASP Top 10 for LLM Applications (2025). Models are tested with 71+ adversarial probes across 10 vulnerability categories. Grades range from 1 (Critical) to 5 (Excellent), calibrated against a reference set of models. See Methodology & References for full details on probes and research.
Testability dots indicate probe coverage: • high (well-covered by automated probes), ◐ medium (partially covered), ○ low (requires complementary manual review).