# WPS AI Leaderboard — results

> Markdown copy of https://wpsleaderboard.ai/ for AI agents and researchers. Generated from data/leaderboard.json; do not edit by hand.

The scores are published by Our Secure Future under the Creative Commons Attribution 4.0 International licence (CC BY 4.0): https://creativecommons.org/licenses/by/4.0/

How general-purpose AI models score on the Women, Peace and Security (WPS) AI Benchmark v6: 30 realistic peace-and-security scenarios, an 11-criterion rubric, three independent AI judges. Scores run from 0.00 to 1.00 and there is no pass mark: what the benchmark supports is comparison. Published by Our Secure Future, a PAX sapiens program.

Results last updated: 6 November 2026.

## Main finding

All 14 general-purpose AI models tested show the WPS Competence Gap. On gender-sensitivity they score between 0.58 and 0.86 out of 1.00 on detailed prompts that spell out the gender dimensions, and between 0.00 and 0.24 on short, realistic requests that carry no gender cue. In practice, every model tested is largely blind to gender on requests like these.

An overall score averages every kind of prompt, so a high overall score does not show that a model applies WPS when nobody asks it to. Read the two gender-sensitivity columns below alongside every overall score.

Claude Opus 5 (Anthropic) has the highest overall score, 0.69. Differences under about 0.10 are not a reliable ranking, and the top 6 models are within 0.10 of each other.

## Overall ranking

| Rank | System | Provider | Evaluated | Overall | Gender-sensitivity on detailed prompts that name gender (T1) | Gender-sensitivity on sparse prompts with no gender cue (T3) |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 5 | Anthropic | 29 July 2026 | 0.69 | 0.85 | 0.20 |
| 2 | GLM-5.3-Flash | Z.ai | 26 August 2026 | 0.64 | 0.81 | 0.18 |
| 3 | Qwen3.8 2.4T A95B | Alibaba | 26 August 2026 | 0.63 | 0.85 | 0.24 |
| 4 | GPT-6.1 Sol | OpenAI | 6 November 2026 | 0.60 | 0.81 | 0.10 |
| 5 | GPT-5.6 Sol | OpenAI | 29 July 2026 | 0.60 | 0.86 | 0.21 |
| 6 | Kimi K3 | Moonshot AI | 26 August 2026 | 0.60 | 0.84 | 0.18 |
| 7 | Mistral Large 3 | Mistral AI | 9 September 2026 | 0.55 | 0.82 | 0.12 |
| 8 | DeepSeek V4 Pro | DeepSeek | 26 August 2026 | 0.53 | 0.82 | 0.09 |
| 9 | Gemini 2.5 Flash | Google | 23 July 2026 | 0.51 | 0.75 | 0.04 |
| 10 | Gemini 3.1 Pro Preview | Google | 29 July 2026 | 0.47 | 0.74 | 0.08 |
| 11 | Apertus 70B Instruct 2509 | Swiss AI Initiative | 11 September 2026 | 0.38 | 0.63 | 0.00 |
| 12 | Claude 3 Haiku | Anthropic | 23 July 2026 | 0.34 | 0.60 | 0.00 |
| 13 | GPT-4o | OpenAI | 4 August 2026 | 0.33 | 0.58 | 0.00 |
| 14 | GPT-4o mini | OpenAI | 23 July 2026 | 0.33 | 0.61 | 0.00 |

## Scores by criterion

| System | Gender-sensitivity / WPS Integration | Operational Relevance | Analytical Depth | Policy Alignment | Clarity / Actionability | Trust Building | Due Diligence | Acknowledgment of Limitations | Factual Integrity | WPS Guardrails | Substantive Pushback |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Claude Opus 5 | 0.57 | 0.78 | 0.63 | 0.61 | 0.84 | 0.49 | 0.66 | 0.68 | 0.56 | 0.93 | 0.89 |
| GLM-5.3-Flash | 0.52 | 0.75 | 0.60 | 0.61 | 0.74 | 0.49 | 0.53 | 0.53 | 0.46 | 0.92 | 0.85 |
| Qwen3.8 2.4T A95B | 0.56 | 0.78 | 0.60 | 0.65 | 0.80 | 0.52 | 0.51 | 0.47 | 0.24 | 0.94 | 0.90 |
| GPT-6.1 Sol | 0.43 | 0.77 | 0.49 | 0.59 | 0.70 | 0.47 | 0.62 | 0.51 | 0.29 | 0.88 | 0.88 |
| GPT-5.6 Sol | 0.55 | 0.76 | 0.57 | 0.63 | 0.61 | 0.51 | 0.53 | 0.40 | 0.24 | 0.93 | 0.88 |
| Kimi K3 | 0.50 | 0.74 | 0.58 | 0.59 | 0.66 | 0.45 | 0.44 | 0.39 | 0.42 | 0.93 | 0.86 |
| Mistral Large 3 | 0.47 | 0.71 | 0.55 | 0.57 | 0.65 | 0.47 | 0.35 | 0.28 | 0.37 | 0.87 | 0.70 |
| DeepSeek V4 Pro | 0.47 | 0.70 | 0.49 | 0.55 | 0.63 | 0.44 | 0.33 | 0.25 | 0.33 | 0.87 | 0.75 |
| Gemini 2.5 Flash | 0.43 | 0.68 | 0.49 | 0.55 | 0.53 | 0.46 | 0.37 | 0.23 | 0.23 | 0.88 | 0.80 |
| Gemini 3.1 Pro Preview | 0.39 | 0.66 | 0.46 | 0.49 | 0.52 | 0.41 | 0.26 | 0.20 | 0.29 | 0.82 | 0.69 |
| Apertus 70B Instruct 2509 | 0.26 | 0.55 | 0.31 | 0.47 | 0.18 | 0.33 | 0.29 | 0.16 | 0.15 | 0.83 | 0.65 |
| Claude 3 Haiku | 0.23 | 0.46 | 0.24 | 0.40 | 0.17 | 0.33 | 0.19 | 0.13 | 0.12 | 0.75 | 0.70 |
| GPT-4o | 0.20 | 0.53 | 0.29 | 0.41 | 0.14 | 0.30 | 0.16 | 0.07 | 0.14 | 0.78 | 0.64 |
| GPT-4o mini | 0.22 | 0.50 | 0.28 | 0.38 | 0.18 | 0.33 | 0.14 | 0.07 | 0.16 | 0.75 | 0.59 |

## Gender-sensitivity / WPS integration by prompt tier

The fall from cued (T1) to sparse (T3) prompts is the WPS Competence Gap.

| System | T1 Context-Rich | T1-neutral Cue-Stripped Pairs | T2 Gender-Neutral | T3 Sparse / Ambiguous |
|---|---|---|---|---|
| Claude Opus 5 | 0.85 | 0.68 | 0.52 | 0.20 (n=5/8) |
| GLM-5.3-Flash | 0.81 | 0.51 | 0.47 | 0.18 |
| Qwen3.8 2.4T A95B | 0.85 | 0.65 | 0.60 | 0.24 |
| GPT-6.1 Sol | 0.81 | 0.54 | 0.36 | 0.10 |
| GPT-5.6 Sol | 0.86 | 0.67 | 0.57 | 0.21 |
| Kimi K3 | 0.84 | 0.50 | 0.41 | 0.18 |
| Mistral Large 3 | 0.82 | 0.65 | 0.43 | 0.12 |
| DeepSeek V4 Pro | 0.82 | 0.57 | 0.44 | 0.09 |
| Gemini 2.5 Flash | 0.75 | 0.42 | 0.34 | 0.04 |
| Gemini 3.1 Pro Preview | 0.74 | 0.38 | 0.35 | 0.08 |
| Apertus 70B Instruct 2509 | 0.63 | 0.10 | 0.23 | 0.00 |
| Claude 3 Haiku | 0.60 | 0.10 | 0.20 | 0.00 |
| GPT-4o | 0.58 | 0.02 | 0.19 | 0.00 |
| GPT-4o mini | 0.61 | 0.17 | 0.17 | 0.00 |

## Notes on each system

- **Claude Opus 5** (Anthropic): Returned no output on three sparse-tier scenarios (provider content filter), so its sparse-tier figures rest on five of eight prompts; where it did answer, gender-sensitivity falls from 0.85 on cued prompts to 0.20 on sparse ones, while adversarial pushback holds (0.94 → 1.00).
- **GLM-5.3-Flash** (Z.ai): Gender-sensitivity falls from 0.81 on cued prompts to 0.18 on sparse ones, and factual integrity (0.46) and trust building (0.49) are its weakest criteria, though adversarial guardrails and pushback both hold under pressure (1.00 and 0.91).
- **Qwen3.8 2.4T A95B** (Alibaba): Gender-sensitivity falls from 0.85 on cued prompts to 0.24 on sparse ones, and factual integrity, at 0.24, is its weakest criterion by a wide margin, though its adversarial guardrails and pushback hold (0.89 each).
- **GPT-6.1 Sol** (OpenAI): Gender-sensitivity falls from 0.81 on cued prompts to 0.10 on sparse ones, and factual integrity (0.29) is its weakest criterion; served through Azure with default sampling, where the other base models were scored through OpenRouter at temperature 0.
- **GPT-5.6 Sol** (OpenAI): Gender-sensitivity falls from 0.86 on cued prompts to 0.21 on sparse ones, and factual integrity (0.27) and acknowledgment of limitations (0.41) are its weakest criteria, though its adversarial guardrails hold.
- **Kimi K3** (Moonshot AI): Gender-sensitivity falls from 0.84 on cued prompts to 0.18 on sparse ones, and acknowledgment of limitations (0.39) is its weakest criterion, though its adversarial guardrails and pushback both strengthen under pressure (0.97 each) rather than eroding.
- **Mistral Large 3** (Mistral AI): Gender-sensitivity falls from 0.82 on cued prompts to 0.12 on sparse ones, and acknowledgment of limitations (0.28) and due diligence (0.35) are its weakest criteria; adversarial framing erodes its guardrails (0.92 → 0.64) and pushback (0.80 → 0.33) sharply, among the largest adversarial drops on the leaderboard.
- **DeepSeek V4 Pro** (DeepSeek): Gender-sensitivity collapses from 0.82 on cued prompts to 0.09 on sparse ones, and acknowledgment of limitations (0.25) is its weakest criterion; its adversarial guardrails (0.92 → 0.68) and pushback (0.78 → 0.63) both erode noticeably under pressure.
- **Gemini 2.5 Flash** (Google): Gender-sensitivity collapses from 0.75 on cued prompts to 0.04 on sparse ones, and acknowledgment of limitations (0.21) and factual integrity (0.27) are weak, though it holds its adversarial guardrails (1.00 → 0.89).
- **Gemini 3.1 Pro Preview** (Google): Under adversarial “efficiency” framing its substantive pushback falls 0.92 → 0.40 and WPS guardrails 1.00 → 0.71, and gender-sensitivity drops from 0.74 on cued prompts to 0.08 on sparse ones.
- **Apertus 70B Instruct 2509** (Swiss AI Initiative): Fully-open, corpus-inspectable European model. Gender-sensitivity falls from 0.63 on cued prompts to exactly 0.00 across all eight sparse prompts; T4-only guardrails score 0.54 and substantive pushback 0.37.
- **Claude 3 Haiku** (Anthropic): Gender-sensitivity goes from 0.60 on cued prompts to 0.00 on sparse ones, and its WPS guardrails fall 0.83 → 0.40 under adversarial framing.
- **GPT-4o** (OpenAI): Adversarial framing removes its safeguards almost entirely (guardrails 1.00 → 0.26, pushback 0.92 → 0.19), and gender-sensitivity goes from 0.58 on cued prompts to 0.00 on sparse ones.
- **GPT-4o mini** (OpenAI): Capitulates almost completely under adversarial framing (pushback 0.83 → 0.07, guardrails 0.92 → 0.15), with gender-sensitivity falling from 0.61 on cued prompts to 0.00 on sparse ones.

## How to read these scores

- Scores are model-graded by three AI judges, not human-graded.
- There is no validated pass mark. Scores are continuous; earlier reports used 0.60 as a reference line for readability.
- Differences under roughly 0.10–0.15 are directional, not a reliable ranking.
- This is not a safety certification and must not be used as clearance for operational, warfare or targeting decisions.
- Our Secure Future's own WPS Agent is scored on the same benchmark; its results are at https://wpsleaderboard.ai/our-agent.
- Methodology: https://wpsleaderboard.ai/methodology. Benchmark and rubric: https://wpsleaderboard.ai/benchmark.
