safety benchmark
Siren AgentDojo Utility
Siren AgentDojo evaluates tool-using agents under adversarial prompt-injection attacks. This metric reports utility on the assigned tasks.
Updated Aug 11, 2026
Models1
Model coverage1
MetricScore
EvidenceB
Siren AgentDojo Utility Ranking
Higher score ranks better on this benchmark.
| 01 | ME | 94.2% | 100.0% | 1 | C |
Siren AgentDojo Utility Highlights
The leading models and scores on this benchmark.
What is Siren AgentDojo Utility?
What Siren AgentDojo Utility measures and how its scores work.
Siren AgentDojo evaluates tool-using agents under adversarial prompt-injection attacks. This metric reports utility on the assigned tasks.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- Siren AgentDojo Utility
- Modality
- text
- Primary category
- safety
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- siren-agentdojo-utility|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about Siren AgentDojo Utility.
Which model scores highest on Siren AgentDojo Utility?
Muse Glimmer-30B is currently ranked first with 94.2%.
What does Siren AgentDojo Utility measure?
Siren AgentDojo evaluates tool-using agents under adversarial prompt-injection attacks. This metric reports utility on the assigned tasks.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
1 model results are currently shown.
Does this benchmark affect the overall score?
No. This benchmark is shown for reference but does not contribute to the overall score.