NEWS · MODELS · #556
Vals AI’s GPT-6 Astra completes multiple long-horizon game milestones and posts big ARC-AGI-3 gains
According to Vals AI and related community runs, GPT-6 Astra reached far-end Minecraft goals (built a Nether portal and gathered end resources before a Creeper destroyed its chest), won Pokemon FireRed in about 18 hours (vs. ~96 hours for GPT‑5.6 Sol), launched a Factorio rocket in ~10 hours, and more, while ARC Prize reported Astra scored ~62.7% on the ARC‑AGI‑3 benchmark (vs. ~7.78% for GPT‑5.6 Sol). ARC Prize and Vals AI attribute the jump to Astra’s ability to form compact symbolic descriptions from observations and turn them into reusable plans while operating through general screen/mouse/keyboard interfaces.
KEY POINTS
- According to Vals AI and related community runs, GPT-6 Astra reached far-end Minecraft goals (built a Nether portal and gathered end resources before a Creeper destroyed its chest), won Pokemon FireRed in about 18 hours (vs.
- ~96 hours for GPT‑5.6 Sol), launched a Factorio rocket in ~10 hours, and more, while ARC Prize reported Astra scored ~62.7% on the ARC‑AGI‑3 benchmark (vs.
- ARC Prize and Vals AI attribute the jump to Astra’s ability to form compact symbolic descriptions from observations and turn them into reusable plans while operating through general screen/mouse/keyboard interfaces.
WHY IT MATTERS
This matters because it indicates a substantial jump in general‑interface, long‑horizon planning and self‑derived symbolic reasoning capabilities, with implications for capability trajectory and agent research priorities.