ARC-AGI reports GPT-6 Astra results under two test harnesses

ARC Prize reported that OpenAI's GPT-6 Astra scored 62.7% on ARC-AGI-3 Semi-Private under its Standard harness and 99.9% under a Provider Adapter harness that preserves opaque reasoning state between requests. The organization said the two setups answer different questions and are not directly equivalent to its controlled human testing. In the Provider Adapter results, Astra used fewer actions than the stated human baseline on 96% of levels. ARC Prize said the benchmark has a tightly bounded, deterministic format and does not establish general intelligence.