GPT 6 Astra Scores 53.6 on Elorian Benchmark, Humans Hit 93.1
Summary
Elorian and Scale AI release Humanity's Sixth Sense, a 522-question visual reasoning benchmark, where GPT 6 Astra scores 53.6 pass@1 against a human 93.1. Reflection says its Beam model, 501 billion parameters with 23 billion active, trails Chinese open weight rivals on benchmarks.
Key Points
- Elorian and Scale AI say HSS spans 288 images and 234 video clips covering 17.6 hours.
- Reflection says Beam pre-trains on 6,144 GB300 NVL72 GPUs for four weeks.
- Jack Clark writes he now expects full recursive self-improvement to be possible in 2027.