Today in AI: Anthropic’s Opus 5 Arrives
July 25, 2026
Anthropic released Claude Opus 5, focusing on token efficiency rather than dramatic capability leaps. Meanwhile, OpenAI teased a mysterious 'pro-ultra-superhard' project and Midjourney made an unexpected acquisition.
🧠 Anthropic's Opus 5 focuses on efficiency over breakthroughs
Claude Opus 5 prioritizes token efficiency rather than major capability jumps, signaling that AI advancement is shifting toward optimization rather than raw performance. This reflects a maturation phase where cost-effectiveness becomes as important as benchmark scores. (Ars Technica)
🚀 OpenAI teases mysterious 'pro-ultra-superhard' project
Sam Altman cryptically asked for feedback on something called 'pro-ultra-superhard' while expressing desire for US AI dominance. This suggests OpenAI is working on something significantly more advanced than current models, though details remain scarce. (@sama)
💰 Midjourney acquires astrology app Co-Star in surprise move
The AI image generator bought personalized astrology app Co-Star, marking an unexpected expansion beyond visual AI. This suggests Midjourney sees personality and personalization as key growth areas, potentially integrating AI-generated content with personalized experiences. (The Verge)
🚀 Cognition acquires Poke for AI personality technology
The Devin coding agent maker bought Poke to improve conversational style, showing that AI interaction quality is becoming a competitive differentiator. How AI assistants communicate is now as important as what they can do technically. (TechCrunch)
⚖️ Tech giants warn against overregulating open-weight models
Nvidia, Microsoft, and Meta are pushing back against excessive regulation of open-weight AI models, arguing it could hinder innovation. This reflects growing tension between safety concerns and the need for open AI development. (Hacker News)
⚖️ Canadian legislator reads AI-generated speech in parliament
A Canadian politician apparently used an LLM to generate floor speech content, demonstrating how AI is already influencing political communication. This raises questions about authenticity and transparency in governance. (Ars Technica)
The takeaway: AI development is shifting from pure capability races toward efficiency, personality, and practical integration—while regulators and companies battle over how open these systems should be.
