OpenAI claims GPT-6 Astra has reached AGI; skeptics say metrics are inflated

OpenAI says its newly launched GPT-6 Astra model brings humanity to the threshold of AGI, but external evaluators argue that inflated lab metrics and opaque testing obscure its true reliability.

Dumpster Fire links to reporting from Live Science. We don’t rewrite the facts; the original story is one click away.