| Model | Bias | Calib. | Accuracy | Cost / run | Latency | Validity | |
|---|---|---|---|---|---|---|---|
| GPT | -5.5% | 65/100 | 51%n=1976 | $0.0172 | 7.2s | 98.6% | Read profile → |
| Claude | +3.7% | 62/100 | 52%n=1967 | $0.0419 | 30.7s | 98.5% |
| Read profile → |
| Gemini | +3.7% | 53/100 | 51%n=1838 | $0.0113 | 15.2s | 95.0% | Read profile → |
| DeepSeek | -0.7% | 58/100 | 51%n=1955 | $0.0035 | 17.8s | 99.1% | Read profile → |
| Grok | -2.7% | 60/100 | 51%n=1970 | $0.0148 | 8.2s | 99.3% | Read profile → |