Source: Veridrop · Tested host: api.tu-zi.com
- Median score
- 93 · At risk
- Tests in total
- 19
- Last test
- Days with tests in the last 90 days
- 12
- Protocol
- Claude, Gemini, OpenAI
Pass rate by check
| Check | Protocol | Passed |
|---|---|---|
| Behaviour pattern check | Claude | 5 of 7 |
| Model field and output stability check | Claude | 6 of 7 |
| Self-reported identity check | Claude | 6 of 7 |
| Streaming consistency check | Claude | 6 of 7 |
| Message ID format check | Claude | 6 of 6 |
| Protocol field check | Claude | 6 of 6 |
| Thinking signature check | Claude | 5 of 7 |
| Usage figure consistency check | Claude | 3 of 7 |
| Basic request check | Gemini | 2 of 2 |
| Function call check | Gemini | 2 of 2 |
| Streaming consistency check | Gemini | 2 of 2 |
| Model field and output stability check | Gemini | 1 of 2 |
| Protocol field check | Gemini | 2 of 2 |
| Structured output check | Gemini | 2 of 2 |
| Usage figure consistency check | Gemini | 1 of 2 |
| Basic request check | OpenAI | 9 of 10 |
| Function call check | OpenAI | 7 of 10 |
| Streaming consistency check | OpenAI | 9 of 10 |
| Model field and output stability check | OpenAI | 8 of 10 |
| Protocol field check | OpenAI | 5 of 10 |
| Structured output check | OpenAI | 8 of 10 |
| Usage figure consistency check | OpenAI | 4 of 10 |
- Counting window
- —
- Score standard deviation
- 26
- Reports with serious issues
- 5 of 19