# The Multivac — Evaluation Report

**Evaluation ID:** EVAL-20260402-162422
**Date:** Apr 02, 2026
**Category:** reasoning
**Question ID:** REASON-012

---

## Question

A disease affects 1 in 10,000 people. A test for the disease is 99% sensitive (true positive rate) and 99.5% specific (true negative rate). A patient tests positive. (1) What is the probability they have the disease? (2) If they test positive twice with independent tests, what is the probability? (3) A doctor says 'You tested positive, so you almost certainly have it.' Critique this reasoning. (4) Design a testing protocol that achieves >95% positive predictive value.

---

## Winner

**DeepSeek V4** (openrouter)
- Winner Score: 8.93
- Matrix Average: 7.92
- Total Judgments: 48

---

## Rankings

| Rank | Model | Provider | Avg Score | Judgments |
|------|-------|----------|-----------|----------|
| 1 | GPT-5.4 | openrouter | 9.54 | 6 |
| 2 | Claude Sonnet 4.6 | openrouter | 9.50 | 6 |
| 3 | Claude Opus 4.6 | openrouter | 9.17 | 6 |
| 4 | DeepSeek V4 | openrouter | 8.93 | 8 |
| 5 | MiMo-V2-Flash | Xiaomi | 8.90 | 2 |
| 6 | GPT-OSS-120B | OpenAI | 8.80 | 3 |
| 7 | MiniMax M2.5 | openrouter | 8.50 | 4 |
| 8 | Gemini 2.5 Flash | openrouter | 6.63 | 4 |
| 9 | Grok 4.20 | openrouter | 6.15 | 3 |
| 10 | Gemini 3.1 Pro | openrouter | 3.12 | 6 |

---

## 10×10 Judgment Matrix

Rows = Judge, Columns = Respondent. Self-judgments excluded (—).

| Judge ↓ / Resp → | DeepSeek V4 | MiMo-V2-Flash | GPT-OSS-120B | Claude Opus | GPT-5.4 | Grok 4.20 | Claude Sonnet | Gemini 2.5 | MiniMax M2.5 | Gemini 3.1 Pro |
|---|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4 | — | · | · | · | · | · | · | 8.1 | 8.1 | · |
| MiMo-V2-Flash | 9.4 | — | 8.6 | 10.0 | 9.8 | 8.0 | 10.0 | 8.6 | 8.3 | 2.6 |
| GPT-OSS-120B | 9.1 | · | — | 8.1 | 9.1 | · | 8.8 | · | · | 3.2 |
| Claude Opus | 7.8 | · | · | — | · | · | 9.4 | 1.3 | · | · |
| GPT-5.4 | 8.6 | · | · | · | — | · | · | · | · | · |
| Grok 4.20 | 8.8 | 8.8 | 8.8 | 8.8 | 8.8 | — | 9.0 | 8.6 | 8.6 | 1.1 |
| Claude Sonnet | 8.4 | · | · | 9.0 | 9.8 | · | — | · | · | 1.6 |
| Gemini 2.5 | 10.0 | 9.0 | 9.0 | 9.4 | 10.0 | 9.0 | 10.0 | — | 9.0 | 8.4 |
| MiniMax M2.5 | 9.4 | · | · | 9.7 | 9.8 | 1.4 | 9.8 | · | — | 1.6 |
| Gemini 3.1 Pro | · | · | · | · | · | · | · | · | · | — |

---

## Methodology

- **10×10 Blind Peer Matrix:** All models answer the same question, then all models judge all responses.
- **5 Criteria:** Correctness, completeness, clarity, depth, usefulness (each scored 1–10).
- **Self-judgments excluded:** Models do not judge their own responses.
- **Weighted Score:** Composite of all 5 criteria.

---

## Citation

The Multivac (2026). Blind Peer Evaluation: REASON-012. app.themultivac.com

## License

Open data. Free to use, share, and build upon. Please cite The Multivac when using this data.

Download raw JSON: https://app.themultivac.com/api/evaluations/EVAL-20260402-162422/results
Full dataset: https://app.themultivac.com/dashboard/export