GPT-6 Astra shows 'step change' in spatial reasoning, solving 7/100 robot tasks vs zero for rival

The Decoder · rss · 2026-09-12

Early benchmarks suggest GPT-6 Astra delivers a "step change" in spatial reasoning, The Decoder reports. On the robotics benchmark StationeryBench, the model completed 7 out of 100 dual-arm robot tasks, while competitor MolmoAct2 couldn't finish a single one. One researcher calls it "a step change in spatial reasoning."

Original post →

More from Embodied

Embodied channel →