Archived v2.1Unresolved leading group
GPT-5.6 Terra Pro
This profile reports GPT-5.6 Terra Pro’s original v2.1 result. It does not claim a timeless culinary rank. The run’s scoring limitations and later exploratory analyses are separate records.
Original overall
95.4
95% item-bootstrap interval
91.7–98.3
Frontier subset
91.3
Original score profile
- Quantities & Scaling
- 100.0
- Conversions
- 100.0
- Food Safety
- 95.0
- Substitutions
- 93.1
- Technique
- 99.1
- Flavour Pairing
- 93.6
- Nutrition
- 90.0
Uncertainty and pairwise evidence
GPT-5.6 Terra Pro belonged to a five-model group from which the run’s paired item-bootstrap rule did not identify one unique leader.
Scoring limitations
Saturated numeric items, high-influence keyword items and semantic constraint-check failures affected the instrument. The profile is retained as historical evidence, not validated culinary ability.
Run 2026-07-v2.1 · openai/gpt-5.6-terra-pro · original artifact unchanged