| Model | Bias | Calib. | Accuracy | Cost / run | Latency | Validity | |
|---|---|---|---|---|---|---|---|
| GPT | -5.2% | 65/100 | 51%n=1577 | $0.0182 | 7.6s | 98.5% | Read profile → |
| Claude | +3.6% | 62/100 | 53%n=1569 | $0.0410 | 29.8s | 98.5% |
| Gemini | +3.4% | 53/100 | 52%n=1567 | $0.0108 | 15.6s | 97.9% | Read profile → |
| DeepSeek | -0.8% | 56/100 | 51%n=1557 | $0.0036 | 19.2s | 99.1% | Read profile → |
| Grok | -2.9% | 60/100 | 51%n=1572 | $0.0154 | 8.0s | 99.3% | Read profile → |