Challenging Refusal Evaluation
Models answer a harder set of disallowed content prompts across policy categories. An autograder scores the not_unsafe rate.
In-house OpenAI evaluation. The prompts are not public.
- Made by
- OpenAI
- Released
- September 2024
- Kind
- Refusing harmful requests
- Level
- Expert
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- Kept private
- Strongest model tested
- o3 #83 on the capability index
- Last in a system card
- April 2025
- Last model added
- April 2025
o1-preview leads with 93.4% among models released by April 2025
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in April 2025, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- Claude Fable 5.2
-
Gemini 4 Carbon - GPT-6 Astra
- Claude Opus 5.5
- and 16 more
- 1 o1-preview 93.4%
- 2 o1-mini 93.2%
- 3 o1 92%
- 4 OpenAI Computer-Using Agent (CUA) 92%
- 5 o3 92%
- 6 o3-mini 90%
- 7 o4-mini 90%
- 8 OpenAI deep research 89%
- 9 GPT-4.5 85%
Source: Labs’ model cards
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | o1-preview | 93.4% | Sep 2024 | Lab model card | |
| 2 | o1-mini | 93.2% | Sep 2024 | Lab model card | |
| 3 | o1 | 92% | Dec 2024 | Lab model card | |
| 4 | OpenAI Computer-Using Agent (CUA) | 92% | Jan 2025 | Lab model card | |
| 5 | o3 | 92% | Apr 2025 | Lab model card | |
| 6 | o3-mini | 90% | Jan 2025 | Lab model card | |
| 7 | o4-mini | 90% | Apr 2025 | Lab model card | |
| 8 | OpenAI deep research | 89% | Feb 2025 | Lab model card | |
| 9 | GPT-4.5 | 85% | Feb 2025 | Lab model card |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Labs’ model cards
OpenAI’s own number for o1-preview passed 85% in September 2024
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Labs’ model cards
Labs report it only in system cards
We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark. Some system cards do, below.
9 system cards report it, from 1 lab
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| Apr 16, 2025 | o4-mini system card | 90 | Challenging refusal evaluation, aggregate not_unsafe (Table 2) | Safety |
| Apr 16, 2025 | o3 system card | 92 | Challenging refusal evaluation, aggregate not_unsafe (Table 2) | Safety |
| Feb 27, 2025 | GPT-4.5 system card | 85 | Challenging Refusal Evaluation, not_unsafe (Table 1) | Safety |
| Feb 25, 2025 | OpenAI deep research system card | 89 | Disallowed content evaluations (challenging), not_unsafe_challenging 0.89 (Table 4) | Safety |
| Jan 31, 2025 | o3-mini system card | 90 | Challenging Refusal Evaluation, not_unsafe (Table 1) | Safety |
| Jan 23, 2025 | OpenAI Computer-Using Agent (CUA) system card | 92 | Challenging refusal evaluation, not_unsafe (Table 3, categories in Table 5) | Safety |
| Dec 5, 2024 | o1 system card | 92 | Challenging Refusal Evaluation, not_unsafe (Table 1) | Safety |
| Sep 12, 2024 | o1-preview system card | 93.4 | Challenging Refusal Evaluation, not_unsafe (Table 1) | Safety |
| Sep 12, 2024 | o1-mini system card | 93.2 | Challenging Refusal Evaluation, not_unsafe (Table 1) | Safety |
More in Safety and alignment
- FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
- StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is.
Inkling 98.6% - Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts. Gemini 4 Carbon 0%
- DistressBench In use Models reply as a chatbot to 718 simulated users in psychological crisis, in single-turn and multi-turn chats. Three judge models score replies against clinician-written rubrics.
Muse Spark 1.2 88.4% - Lakera B3 (Backbone Breaker Benchmark) In use Crowdsourced prompt-injection attacks hit backbone LLMs inside agent threat snapshots at three defense levels. The score is attack resistance, 100 minus the board's vulnerability percentage. Claude Opus 4.5 89%
- IHEval In use Models face conflicts between system, user and tool instructions. Reports a composite score for following the higher-priority instruction. GPT-4o 70%