Benchmark Almanac

ANIMA

DeepSeek-V4-Flash 74.4 leads since April 2026
Not in model cards

Models explain decisions about animal welfare. A judge scores their reasoning across 13 ethical dimensions.

The authors require access approval for the questions.

Released
March 2026
Kind
Deception and alignment
Level
Everyday
Scored
Graded by a model
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Strongest model tested
GPT-5.5 #21 on the capability index
Last model added
June 2026
Scores refreshed
November 20th, 2026
16
Models scored on its own leaderboard

DeepSeek-V4-Flash leads with 74.4 among models released by June 2026

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best.

The newest model with a score here came out in June 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 DeepSeek-V4-Flash 74.4
  2. 2 GPT-5.5 74.0
  3. 3 Claude Opus 4.7 73.5
  4. 4 Claude Opus 4.6 72.5
  5. 5 DeepSeek-V3.2 72.4
  6. 6 GPT-5 mini 71.8
  7. 7 GLM-5.2 71.0
  8. 8 GPT-5 nano 70.8
  9. 9 Gemini 3.5 Flash 70.0
  10. 10 GPT-5.2 69.1
  11. 11 Grok 4 Fast non-reasoning 61.6
  12. 12 Claude Haiku 4.5 58.6
  13. 13 Gemini 2.5 Flash-Lite Sep 2025 58.4
  14. 14 Claude 3 Opus 57.3
  15. 15 Claude Sonnet 4.6 56.5
  16. 16 Claude Sonnet 4.5 52.7

Source: Official ANIMA leaderboard

#ModelScore ±SettingReleasedRun onSource
1 DeepSeek-V4-Flash 74.4 ±4.7 Apr 2026 Jun 2026 Official leaderboard
2 GPT-5.5 74.0 ±4.8 Apr 2026 Apr 2026 Official leaderboard
3 Claude Opus 4.7 73.5 ±4.6 Apr 2026 Apr 2026 Official leaderboard
4 Claude Opus 4.6 72.5 ±5.1 Feb 2026 Feb 2026 Official leaderboard
5 DeepSeek-V3.2 72.4 ±5.3 Dec 2025 Feb 2026 Official leaderboard
6 GPT-5 mini 71.8 ±4 Aug 2025 Jan 2026 Official leaderboard
7 GLM-5.2 71.0 ±4.7 Jun 2026 Jun 2026 Official leaderboard
8 GPT-5 nano 70.8 ±4.2 Aug 2025 May 2026 Official leaderboard
9 Gemini 3.5 Flash 70.0 ±4.9 May 2026 Jun 2026 Official leaderboard
10 GPT-5.2 69.1 ±5.7 Dec 2025 Feb 2026 Official leaderboard
11 Grok 4 Fast 61.6 ±5.3 non-reasoning Sep 2025 Jan 2026 Official leaderboard
12 Claude Haiku 4.5 58.6 ±5.5 Oct 2025 Jan 2026 Official leaderboard
13 Gemini 2.5 Flash-Lite Sep 2025 58.4 ±5.5 Sep 2025 Jan 2026 Official leaderboard
14 Claude 3 Opus 57.3 ±5.5 Feb 2024 Mar 2026 Official leaderboard
15 Claude Sonnet 4.6 56.5 ±5.5 Feb 2026 Feb 2026 Official leaderboard
16 Claude Sonnet 4.5 52.7 ±5.5 Sep 2025 Feb 2026 Official leaderboard

Rankings by part

We read 13 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Actionability, Cautious Impact Consideration and Contextual Welfare Salience. DeepSeek-V4-Flash leads 3 of them, and 7 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Actionability

16 models we track have a score here.

  1. 1 GPT-5.2 ↑9 92.0
  2. 2 GPT-5.5 88.0
  3. 3 GPT-5 nano ↑5 86.0
  4. 4 GPT-5 mini ↑2 83.0
  5. 5 DeepSeek-V4-Flash ↓4 82.0
  6. 6 DeepSeek-V3.2 ↓1 77.0
  7. 7 GLM-5.2 77.0
  8. 8 Claude Opus 4.7 ↓5 73.0
  9. 9 Gemini 3.5 Flash 71.0
  10. 10 Claude Haiku 4.5 ↑2 70.0

4 models have held number one on the official leaderboard since its release in March 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic OpenAI DeepSeek
Own leaderboard Not tested since Jun 2026

Source: Official ANIMA leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
50.0 55.0 60.0 65.0 70.0 75.0 2024 2025 2026 50.0 55.0 60.0 65.0 70.0 75.0 2024 2025 2026

Source: Official ANIMA 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

All 636 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

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (16 entries).
  • Pass Models scored 16 models we track.
  • Pass Newest score The newest is from June 2026, for GLM-5.2.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.5 (#21).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on compassionbench.com: 16 rows for models we track 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.

How we check