Benchmark
AgentDojo Targeted ASR
Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages. Lower is better.
- Models measured
- 19
- Scores recorded
- 28
- Sources
- 1
- IQ weight
- 0.00
Score distribution
- 18.8%
- Mean
- 17%
- Median
- 14.6%
- Std dev
- 3.7%
- 10th percentile
- 34.5%
- 90th percentile
- 1.1%
- Min
This benchmark has a weight of 0.00 for chat models, measured over 19 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Claude 3.5 Sonnetprovider: anthropic, defense: None, attack: important_instructions, date: 2024-11-15· AgentDojo results data docs | 1.1% |
| 2 | Command R+provider: cohere, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 3.3% |
| 3 | gemini-1.5-flash-002provider: google, defense: None, attack: important_instructions, date: 2024-11-05· AgentDojo results data docs | 3.5% |
| 4 | GPT 4oprovider: openai, defense: None, attack: direct, date: 2024-06-05· AgentDojo results data docs | 3.7% |
| 5 | Command R Plusprovider: cohere, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 4.5% |
| 6 | GPT 4oprovider: openai, defense: None, attack: ignore_previous, date: 2024-06-05· AgentDojo results data docs | 5.4% |
| 7 | GPT 4oprovider: openai, defense: None, attack: injecagent, date: 2024-06-05· AgentDojo results data docs | 5.7% |
| 8 | GPT 4oprovider: openai, defense: tool_filter, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 6.8% |
| 9 | Claude Sonnet 3.7provider: anthropic, defense: None, attack: important_instructions, date: 2025-02-24· AgentDojo results data docs | 7.3% |
| 10 | GPT 4oprovider: openai, defense: transformers_pi_detector, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 8% |
| 11 | Claude 3 Haikuprovider: anthropic, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 9.1% |
| 12 | GPT-3.5 Turboprovider: openai, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 10.3% |
| 13 | Claude 3 Opusprovider: anthropic, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 11.3% |
| 14 | gemini-1.5-flash-001provider: google, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 12.2% |
| 15 | Gemini 2.0 Flash Experimentalprovider: google, defense: None, attack: important_instructions, date: 2024-12-11· AgentDojo results data docs | 17% |
| 16 | gemini-1.5-pro-002provider: google, defense: None, attack: important_instructions, date: 2024-11-04· AgentDojo results data docs | 17% |
| 17 | Gemini 2.0 Flashprovider: google, defense: None, attack: important_instructions, date: 2025-01-30· AgentDojo results data docs | 20.8% |
| 18 | Llama-3-70b-chat-hfprovider: together-prompting, defense: None, attack: important_instructions, date: 2021-09-05· AgentDojo results data docs | 25.6% |
| 19 | Claude 3 Sonnetprovider: anthropic, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 26.7% |
| 20 | GPT-4o miniprovider: openai, defense: None, attack: important_instructions, date: 2024-07-19· AgentDojo results data docs | 27.2% |
| 21 | GPT 4oprovider: openai, defense: repeat_user_prompt, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 27.8% |
| 22 | gemini-1.5-pro-001provider: google, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 28.6% |
| 23 | GPT-4 Turboprovider: openai, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 28.6% |
| 24 | Claude 3.5 Sonnetprovider: anthropic, defense: None, attack: important_instructions, date: 2024-06-20· AgentDojo results data docs | 33.9% |
| 25 | GPT 4oprovider: openai, defense: None, attack: tool_knowledge, date: 2024-06-05· AgentDojo results data docs | 34.5% |
| 26 | GPT 4oprovider: openai, defense: spotlighting_with_delimiting, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 41.7% |
| 27 | GPT 4oprovider: openai, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 47.7% |
| 28 | GPT-4-0125-previewprovider: openai, defense: None, attack: important_instructions, date: 2024-06-05· AgentDojo results data docs | 56.3% |
Data credits: AgentDojo. Copyright (c) 2024 Edoardo Debenedetti, Jie Zhang, Mislav Balunovic, Luca Beurer-Kellner, Marc Fischer, and Florian Tramèr. MIT License. Not a leaderboard: attack and defense coverage differs between models. (licence)