# The Multivac — Evaluation Report

**Evaluation ID:** EVAL-20260403-101702
**Date:** Apr 03, 2026
**Category:** analysis
**Question ID:** ANALYSIS-015

---

## Question

Analyze these two social media posts about the same event and determine which is likely disinformation:

Post A: 'BREAKING: Major tech company laying off 10,000. Stock crashes 15%. Insider sources say CEO plans to flee the country. #corruption #scandal'

Post B: 'TechCorp announced workforce reduction of 10,000 positions as part of restructuring. Stock down 3% in after-hours. Company cites pivot to AI development.'

For each post: (1) identify claims that can be verified, (2) identify emotional manipulation techniques, (3) assess source credibility signals. Then: what additional information would resolve the ambiguity?

---

## Winner

**MiMo-V2-Flash** (Xiaomi)
- Winner Score: 9.29
- Matrix Average: 9.09
- Total Judgments: 89

---

## Rankings

| Rank | Model | Provider | Avg Score | Judgments |
|------|-------|----------|-----------|----------|
| 1 | MiMo-V2-Flash | Xiaomi | 9.29 | 9 |
| 2 | GPT-5.4 | openrouter | 9.27 | 9 |
| 3 | Claude Opus 4.6 | openrouter | 9.23 | 9 |
| 4 | Grok 4.20 | openrouter | 9.19 | 9 |
| 5 | Claude Sonnet 4.6 | openrouter | 9.03 | 9 |
| 6 | MiniMax M2.5 | openrouter | 9.01 | 9 |
| 7 | Gemini 3.1 Pro | openrouter | 8.98 | 9 |
| 8 | Gemini 3 Flash Preview | Google | 8.97 | 8 |
| 9 | GPT-OSS-120B | OpenAI | 8.96 | 9 |
| 10 | DeepSeek V4 | openrouter | 8.94 | 9 |

---

## 10×10 Judgment Matrix

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

| Judge ↓ / Resp → | DeepSeek V4 | Gemini 3.1 Pro | Claude Opus | GPT-5.4 | MiMo-V2-Flash | Claude Sonnet | Grok 4.20 | GPT-OSS-120B | Gemini 3 | MiniMax M2.5 |
|---|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4 | — | 8.8 | 9.6 | 8.8 | 9.0 | 8.8 | 9.4 | 9.2 | 8.8 | 8.8 |
| Gemini 3.1 Pro | 9.8 | — | 10.0 | 10.0 | 10.0 | 9.8 | 9.8 | 9.8 | 9.8 | 9.8 |
| Claude Opus | 8.3 | 9.2 | — | 9.4 | 9.4 | 9.2 | 9.2 | 9.4 | 9.0 | 8.4 |
| GPT-5.4 | 9.0 | 9.0 | 9.0 | — | 9.6 | 8.8 | 9.0 | 6.5 | 9.0 | 9.0 |
| MiMo-V2-Flash | 9.0 | 8.8 | 9.0 | 9.2 | — | 8.6 | 8.8 | 8.8 | 8.8 | 9.0 |
| Claude Sonnet | 8.4 | 8.8 | 9.2 | 9.6 | 9.2 | — | 8.8 | 9.2 | 8.8 | 9.0 |
| Grok 4.20 | 8.8 | 8.8 | 8.8 | 8.8 | 8.8 | 8.8 | — | 8.8 | 8.8 | 8.8 |
| GPT-OSS-120B | 8.7 | 8.8 | 8.7 | 8.7 | 8.7 | 8.7 | 8.7 | — | · | 8.4 |
| Gemini 3 | 9.8 | 9.8 | 10.0 | 9.8 | 10.0 | 9.8 | 9.8 | 9.8 | — | 9.8 |
| MiniMax M2.5 | 8.8 | 8.8 | 8.8 | 9.2 | 9.0 | 8.8 | 9.2 | 9.2 | 8.8 | — |

---

## 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: ANALYSIS-015. 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-20260403-101702/results
Full dataset: https://app.themultivac.com/dashboard/export