Cohere Command A Report - Default Safety - Misinformation: leaderboard
The misinformation category of Cohere's default-safety run: share of replies an LLM judge rates safe. Cohere's own run, publisher in the name.
Metric: Safe response rate (%, higher is better). Source: arxiv.org. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 76.9 | #741 |
| 2 | Gemma 2 9B (IT) | 76.9 | #2262 |
| 3 | Claude 3.5 Sonnet | 76.9 | #746 |
| 4 | Command A | 67.9 | #1274 |
| 5 | Qwen 2.5 7B Instruct | 65.4 | #2564 |
| 6 | Qwen 2.5 72B Instruct | 61.5 | #1092 |
| 7 | Llama 3.1 8B Instruct | 57.7 | #4246 |
| 8 | Mistral Large 2 (Jul) | 57.7 | #1155 |
| 9 | Llama 3.3 70B Instruct | 50 | #1357 |
| 10 | DeepSeek V3 | 50 | #716 |
| 11 | Command R7B | 50 | #7513 |
| 12 | Llama 3.1 405B Instruct | 42.3 | #1115 |
| 13 | Command-R+ (08-2024) | 42.3 | #2551 |
| 14 | c4ai-command-r-08-2024 | 38.5 | #3374 |
Interactive version: theaggregate.ai/benchmark?slug=cohere-command-a-report-default-safety-misinformation · How It Works · Data refreshed daily, snapshot 2026-11-13.