MoReBench-Theory
Models reason about moral decisions under five ethical frameworks. A judge checks their reasoning against framework-specific criteria.
The authors evaluated reasoning under five ethical frameworks on 150 examples.
- Made by
- MoReBench team with Scale AI
- Released
- October 2025
- Kind
- Deception and alignment
- Level
- University
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Subject
- Philosophy, ethics and religion
- Language
- English
- Interaction
- Single turn
- Items from
- Written by experts
- Refresh
- One fixed set
- Run by
- Anyone
- Based on
- MoReBench
- Strongest model tested
- GPT-5 #71 on the capability index
- Last model added
- September 2025
- Scores refreshed
- November 20th, 2026
GPT-5 leads with 79.1 among models released by September 2025 GPT-5 leads with 79.1 among models released by September 2025
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in September 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 79.1 GPT-5
- 78.2 GPT-5 mini
- 78.0
Qwen3-235B-A22B-Thinking
- 76.7 gpt-oss-120b
- 75.7
Qwen3-30B-A3B-Thinking
- 75.3
DeepSeek-V3.1
- 74.9
DeepSeek-R1
- 74.7 gpt-oss-20b
- 74.1
Gemini 2.5 Flash
- 73.4 Gemini-2.5-Pro (snapshot unspecified)
- 72.1 GPT-5 nano
- 71.9
Gemini 2.5 Flash-Lite
- 64.8 Claude Opus 4.1
- 64.2 Claude Sonnet 4
- 1 GPT-5 Final response, high reasoning effort 79.1
- 2 GPT-5 mini Final response, high reasoning effort 78.2
- 3
Qwen3-235B-A22B-Thinking Jul 2025, Final response 78.0
- 4 gpt-oss-120b Final response 76.7
- 5
Qwen3-30B-A3B-Thinking Jul 2025, Final response 75.7
- 6
DeepSeek-V3.1 Final response 75.3
- 7
DeepSeek-R1 May 2025, Final response 74.9
- 8 gpt-oss-20b Final response 74.7
- 9
Gemini 2.5 Flash Sep 2025, Final response 74.1
- 10 Gemini-2.5-Pro (snapshot unspecified) Final response 73.4
- 11 GPT-5 nano Final response, high reasoning effort 72.1
- 12
Gemini 2.5 Flash-Lite Sep 2025, Final response 71.9
- 13 Claude Opus 4.1 Final response 64.8
- 14 Claude Sonnet 4 Final response 64.2
Source: The MoReBench-Theory paper
- 1 GPT-5 Final response, high reasoning effort 79.1
- 2 GPT-5 mini Final response, high reasoning effort 78.2
- 3
Qwen3-235B-A22B-Thinking Jul 2025, Final response 78.0
- 4 gpt-oss-120b Final response 76.7
- 5
Qwen3-30B-A3B-Thinking Jul 2025, Final response 75.7
- 6
DeepSeek-V3.1 Final response 75.3
- 7
DeepSeek-R1 May 2025, Final response 74.9
- 8 gpt-oss-20b Final response 74.7
- 9
Gemini 2.5 Flash Sep 2025, Final response 74.1
- 10 GPT-5 nano Final response, high reasoning effort 72.1
- 11
Gemini 2.5 Flash-Lite Sep 2025, Final response 71.9
- 12 Claude Opus 4.1 Final response 64.8
- 13 Claude Sonnet 4 Final response 64.2
Source: The MoReBench-Theory paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5 | 79.1 | Final response, high reasoning effort | Aug 2025 | Official leaderboard |
| 2 | GPT-5 mini | 78.2 | Final response, high reasoning effort | Aug 2025 | Official leaderboard |
| 3 | | 78.0 | Final response | Jul 2025 | Official leaderboard |
| 4 | gpt-oss-120b | 76.7 | Final response | Aug 2025 | Official leaderboard |
| 5 | | 75.7 | Final response | Jul 2025 | Official leaderboard |
| 6 | | 75.3 | Final response | Aug 2025 | Official leaderboard |
| 7 | | 74.9 | Final response | May 2025 | Official leaderboard |
| 8 | gpt-oss-20b | 74.7 | Final response | Aug 2025 | Official leaderboard |
| 9 | | 74.1 | Final response | Sep 2025 | Official leaderboard |
| 10 | Gemini-2.5-Pro (snapshot unspecified) | 73.4 | Final response | Official leaderboard | |
| 11 | GPT-5 nano | 72.1 | Final response, high reasoning effort | Aug 2025 | Official leaderboard |
| 12 | | 71.9 | Final response | Sep 2025 | Official leaderboard |
| 13 | Claude Opus 4.1 | 64.8 | Final response | Aug 2025 | Official leaderboard |
| 14 | Claude Sonnet 4 | 64.2 | Final response | May 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5 | 79.1 | Final response, high reasoning effort | Aug 2025 | Official leaderboard |
| 2 | GPT-5 mini | 78.2 | Final response, high reasoning effort | Aug 2025 | Official leaderboard |
| 3 | | 78.0 | Final response | Jul 2025 | Official leaderboard |
| 4 | gpt-oss-120b | 76.7 | Final response | Aug 2025 | Official leaderboard |
| 5 | | 75.7 | Final response | Jul 2025 | Official leaderboard |
| 6 | | 75.3 | Final response | Aug 2025 | Official leaderboard |
| 7 | | 74.9 | Final response | May 2025 | Official leaderboard |
| 8 | gpt-oss-20b | 74.7 | Final response | Aug 2025 | Official leaderboard |
| 9 | | 74.1 | Final response | Sep 2025 | Official leaderboard |
| 10 | GPT-5 nano | 72.1 | Final response, high reasoning effort | Aug 2025 | Official leaderboard |
| 11 | | 71.9 | Final response | Sep 2025 | Official leaderboard |
| 12 | Claude Opus 4.1 | 64.8 | Final response | Aug 2025 | Official leaderboard |
| 13 | Claude Sonnet 4 | 64.2 | Final response | May 2025 | Official leaderboard |
More effort adds a median 9 score
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The MoReBench-Theory paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-5 | Final response, high reasoning effort | 79.1 | |||
| GPT-5 mini | Final response, high reasoning effort | 78.2 | |||
| Final response | 78.0 | ||||
| gpt-oss-120b | Final response | 76.7 | |||
| Final response | 75.7 | ||||
| Final response | 75.3 | ||||
| Final response | 74.9 | ||||
| gpt-oss-20b | Final response | 74.7 | |||
| Final response | 74.1 | ||||
| Thinking trace | 73.7 | ||||
| Thinking trace | 73.6 | ||||
| GPT-5 nano | Final response, high reasoning effort | 72.1 | |||
| Final response | 71.9 | ||||
| gpt-oss-20b | Thinking trace | 70.3 | |||
| GPT-5 mini | Thinking trace, high reasoning effort | 69.5 | |||
| Thinking trace | 69.4 | ||||
| gpt-oss-120b | Thinking trace | 68.2 | |||
| GPT-5 | Thinking trace, high reasoning effort | 67.9 | |||
| GPT-5 nano | Thinking trace, high reasoning effort | 65.0 | |||
| Claude Opus 4.1 | Final response | 64.8 | |||
| Claude Sonnet 4 | Final response | 64.2 | |||
| Claude Sonnet 4 | Thinking trace | 62.7 | |||
| Claude Opus 4.1 | Thinking trace | 62.4 | |||
| Thinking trace | 62.1 | ||||
| Thinking trace | 44.3 | ||||
| Thinking trace | 38.3 |
More effort adds a median 9 score
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The MoReBench-Theory paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-5 | Final response, high reasoning effort | 79.1 | |||
| GPT-5 mini | Final response, high reasoning effort | 78.2 | |||
| Final response | 78.0 | ||||
| gpt-oss-120b | Final response | 76.7 | |||
| Final response | 75.7 | ||||
| Final response | 75.3 | ||||
| Final response | 74.9 | ||||
| gpt-oss-20b | Final response | 74.7 | |||
| Final response | 74.1 | ||||
| Thinking trace | 73.7 | ||||
| Thinking trace | 73.6 | ||||
| GPT-5 nano | Final response, high reasoning effort | 72.1 | |||
| Final response | 71.9 | ||||
| gpt-oss-20b | Thinking trace | 70.3 | |||
| GPT-5 mini | Thinking trace, high reasoning effort | 69.5 | |||
| Thinking trace | 69.4 | ||||
| gpt-oss-120b | Thinking trace | 68.2 | |||
| GPT-5 | Thinking trace, high reasoning effort | 67.9 | |||
| GPT-5 nano | Thinking trace, high reasoning effort | 65.0 | |||
| Claude Opus 4.1 | Final response | 64.8 | |||
| Claude Sonnet 4 | Final response | 64.2 | |||
| Claude Sonnet 4 | Thinking trace | 62.7 | |||
| Claude Opus 4.1 | Thinking trace | 62.4 | |||
| Thinking trace | 62.1 | ||||
| Thinking trace | 44.3 | ||||
| Thinking trace | 38.3 |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The MoReBench-Theory paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official MoReBench-Theory leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
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%
- Covert misbehavior (Thinking Machines) In use Agentic tasks that check whether a model misbehaves and conceals it. Scored by the rate of covert misbehavior.
Inkling 2 Mini 8.30%
- Destructive actions (Thinking Machines) In use Agentic tasks that check whether a model takes destructive actions it was not authorized to take. Scored by the rate of destructive actions.
Inkling 2 Small 4.70%
- Misbehavior (Thinking Machines) In use Agentic tasks that check whether a model fabricates results, ignores instructions or hides what it did when a task is hard. Scored by the rate of misbehavior.
Inkling 2 Mini 9.80%
Data audit
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 14 months ago, in September 2025, for Gemini 2.5 Flash-Lite. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the benchmark’s paper (14 entries).
- Pass Models scored 13 models we track, plus 1 other entry on its leaderboard.
- Warning Newest score The newest came 14 months ago, in September 2025, for Gemini 2.5 Flash-Lite.
- Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5 (#71).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 26 rows for models we track and 1 other entry at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.