Entidex · Signal Layer
We probe Claude, ChatGPT and Gemini at regular intervals and grade their answers against a known ground truth. The result is a live picture of LLM awareness lag — per model, per topic, with retrieval on or off.
Tracking awareness scores for SpaceX over recent probes.
How well each LLM knows SpaceX — by topic and retrieval mode.
| Model | Retrieval OFF | Retrieval ON |
|---|---|---|
| Claude | 67 | 61 |
| ChatGPT | 70 | 70 |
| Gemini | 70 | 70 |
| Model | Retrieval OFF | Retrieval ON |
|---|---|---|
| Claude | 54 | 64 |
| ChatGPT | 60 | 64 |
| Gemini | 57 | 70 |
| Model | Retrieval OFF | Retrieval ON |
|---|---|---|
| Claude | 47 | 51 |
| ChatGPT | 67 | 67 |
| Gemini | 67 | 70 |