GPT-6 Astra scores 62.7% or 99.9% on ARC-AGI-3

GPT-6 Astra can look like it nearly aced ARC-AGI-3, or like it scored 62.7 percent, depending on which test harness you count. Both numbers come from the same model and the same foundation writeup.

The ARC Prize Foundation published an independent benchmark post on 3 September 2026, OpenAI's GPT-6 Astra on ARC-AGI-3. On ARC-AGI-3 Semi-Private, Astra scored 62.7% under the Standard harness and 99.9% under the Provider Adapter harness. The foundation says saturating ARC-AGI-3 is not proof of AGI, and that it will publish both harness results going forward.

The provider-neutral run is the number that does not travel with most launch headlines. Astra (max) scored 62.7% on the Standard harness, where the model carries only the notes it chooses to keep, at a reported cost of about $26,098.

The 99.9% figure is Astra (high) on the Provider Adapter harness, which preserves opaque reasoning state between requests and uses compaction, at about $18,817. Across 167 paired solves, Provider Adapter runs were about 3.66x faster by elapsed time and used 49% fewer total tokens. Under that harness, Astra (max) used fewer actions than the human baseline on 96.0% of levels and 51.7% fewer actions per level on average.

OpenAI's own GPT-6 Astra launch post went up the same day. Fortune later said an embargo draft listed ARC-AGI-3 at 98.6%, the live post showed 99.99%, and other published metrics moved after launch, including a hallucination rate that went from 4.2% to 2% and back to 4.2%. OpenAI blamed a CMS bug and then an outage for a pull-and-republish, and told Fortune that harness, reasoning level, and other factors inform evals.

ARC Prize also notes that Astra built compact algebraic and symbolic world models, plus domain-specific shorthand in its notes. That is a replay observation, not a claim the model generalized past the bench's closed-ended games. The same class of document showed up when Claude formalized Fermat in Lean: a research writeup, not a verdict.

The two scores that matter sit next to each other. 62.7% at about $26,098 is the apples-to-apples run. 99.9% at about $18,817 is the provider-tuned one.

Some offers on this page may be paid placements or contain affiliate links.