# HealthBench (length-adjusted) → ARC-AGI-1

Calibration · https://benchgap.net/calibration/healthbenchlengthadjusted/current/to/arcagi1/current · data retrieved 2026-11-18

ARC-AGI-1 is estimated from HealthBench (length-adjusted) with a Michaelis–Menten curve fitted on 6 models measured on both: y = 2.0000·x / (0.62861 + x), R² = 0.36, cross-validated error 4.4 pp. It is used for 1 estimate.

| Estimated model | HealthBench (length-adjusted) | ARC-AGI-1 | Source |
|---|---|---|---|
| [GPT-6.1 Terra](https://benchgap.net/model/gpt-6-1-terra.md) | 55.1% | 93.4% | estimated ± 4.4 pp, low confidence |

## Fitted on

| Model | HealthBench (length-adjusted) | ARC-AGI-1 |
|---|---|---|
| [Claude Opus 5](https://benchgap.net/model/claude-opus-5.md) | 57.8% | 97.5% |
| [Claude Opus 5.5](https://benchgap.net/model/claude-opus-5-5.md) | 60.6% | 97.5% |
| [GPT-6.1 Sol](https://benchgap.net/model/gpt-6-1-sol.md) | 58.5% | 96.5% |
| [GPT-6 Astra](https://benchgap.net/model/gpt-6-astra.md) | 58.3% | 97.5% |
| [GPT-6 Luna](https://benchgap.net/model/gpt-6-luna.md) | 54.5% | 86.7% |
| [GPT-6 Sol](https://benchgap.net/model/gpt-6-sol.md) | 53.2% | 95.5% |
