Accuracy report
Every substantive Ozmyra answer is re-read by a model from a different AI company before you rely on it. In the last 30 days: 27 answers checked (96.4% of answers that make checkable claims); a problem was flagged in 14.8% of them; 16.7% of checked citations supported the answer; 21.4% of checked claims were backed by a quote from their source.
A flag means the reviewing model raised a concern, not a confirmed error.
| Model | Maker | Answers checked | Flagged |
|---|---|---|---|
| Claude Sonnet 5.5 | Anthropic | 12 | — |
| Claude Haiku 5.5 | Anthropic | 7 | — |
| Claude Opus 5.5 | Anthropic | 4 | — |
| Gemini Flash Lite | 3 | — | |
| Grok 4.20 Reasoning | xAI | 1 | — |