According to TheRundownAI, OpenAI’s internal model produced 722 math papers, tackling ~4,000 problems with ~3 hours of ChatGPT Pro compute each.

According to @OpenAI, an internal frontier model produced new math results, released with IAS advisory input and posted at github.com/openai/math.

According to @claudeai, Claude Opus 5 approaches Fable 5 frontier intelligence at half the price, signaling aggressive value in top-tier models.

According to @OpenAI, GPT-5.6 Sol launches in limited preview with Terra for balanced tasks and Luna for fast, affordable high-volume work.

According to SawyerMerritt, S&P assigns SpaceX BBB, citing strong launch, growing connectivity, and risky AI bets with heavy capex and unclear monetization.

According to SawyerMerritt, SpaceXAI’s Palo Alto HQ hosts teams building Grok, leveraging Silicon Valley to recruit elite AI researchers.