OpenAI’s amazing — but vastly oversold — new model Astra

OpenAI 新模型 Astra 数学表现出色，但被过度吹捧

Gary Marcus argues that while OpenAI's internal model Astra solved 10 open problems in math and theoretical CS at ~$2,000, many are committing the fallacy of composition—treating math prowess as proof of imminent AGI. Expertise in one domain doesn't guarantee general competence, and there's no evidence yet that Astra performs reliably on reasoning, writing, or real-world tasks outside math.

Why it matters: Gary Marcus's critique of OpenAI's internal Astra model is a high-signal event: named entities, concrete data (10 open problems, ~$2,000 cost), and a clear argument (fallacy of composition). It provides a discussable analytical frame, not just sentiment. Score isn't higher bec...