| Model | Bias | Calib. | Accuracy | Cost / run | Latency | Validity | |
|---|---|---|---|---|---|---|---|
| GPT | +7.1% | 69/100 | 52%n=788 | $0.0185 | 7.8s | 99.0% | Read profile → |
| Claude | +14.4% | 60/100 | 54%n=783 | $0.0415 | 30.1s | 98.4% |
| Gemini | +14.6% | 38/100 | 52%n=786 | $0.0110 | 15.7s | 97.7% | Read profile → |
| DeepSeek | +14.6% | 45/100 | 52%n=776 | $0.0037 | 19.7s | 98.9% | Read profile → |
| Grok | +12.8% | 54/100 | 52%n=787 | $0.0156 | 8.1s | 99.2% | Read profile → |