OpenAI has reportedly launched GPT-6 Astra Pro, a new iteration of its ChatGPT model, which allows multiple AI agents to operate simultaneously in response to a single user prompt. Separately, Anthropic has granted internal access to its METR evaluation team, giving them employee-level privileges and the ability to publish findings without prior approval. AI
IMPACT The reported launch of GPT-6 Astra Pro suggests advancements in agentic AI capabilities, while Anthropic's move indicates a potential shift in internal evaluation and transparency practices.
RANK_REASON The cluster reports on a new model release (GPT-6 Astra Pro) from a frontier lab (OpenAI), alongside a significant internal operational change at another frontier lab (Anthropic). [lever_c_demoted from frontier_release: ic=2 ai=1.0]
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 3 sources. How we write summaries →