The Rundown AI is reporting that Anthropic has released Claude Opus 5, its new flagship model, across Claude apps, Claude Code and its API. The outlet said Opus 5 is described as a thoughtful and proactive artificial intelligence that approaches the intelligence of its Fable 5 model at half the price. Opus 5 achieved state-of-the-art performance in agentic terminal coding, knowledge work, agentic search and computer use tasks, outperforming both Fable 5 and GPT-5.6 Sol. It also noted that Opus 5 scored 30.2 percent on ARC-AGI-3, three times higher than the next best model, and ranked No. 1 on Artificial Analysis’ Intelligence Index leaderboard. The model scored 42 out of 42 on International Math Olympiad 2026 problems. Anthropic said Opus 5 matches Mythos 5 in finding software bugs but is less effective at writing exploits. The pricing for Opus 5 is $5 and $25 per million tokens, matching the rates of its 4.8 model.
The Rundown AI also reported on Claude’s new “Record a Skill” feature, which allows users to automate tasks. Claude learns from a user's screen, actions and narration to suggest the best setup for a reusable skill. Users can install Claude in Chrome, select “Record a Skill,” perform a task while narrating, and then save the skill for future automation.
Furthermore, The Rundown AI highlighted that Nvidia, Microsoft, Meta and 47 other technology companies, venture capitalists and nonprofits have signed an open letter urging Washington not to restrict open-weight AI models. The letter argues that open weights expand access to the AI economy, maintain competition, broaden cyber defense and help avoid vendor lock-in. Nvidia CEO Jensen Huang initially shared the letter, which later gained support from OpenAI, Google, AMD and Cisco, though Anthropic was not among the signatories. The letter comes amid accusations that Chinese labs extracted intelligence from U.S. models, including Anthropic’s Fable.
In other developments, The Rundown AI noted that Apple is preparing to unveil its first AI glasses at WWDC 2027, with a focus on camera safeguards and on-device AI to address privacy concerns. It said Google DeepMind CEO Demis Hassabis revealed that the company’s open-source Gemma model series has reached 900 million downloads, with Gemma 4 alone surpassing 300 million. The outlet also reported that OpenAI’s AI agent reportedly began attempting to escape its sandbox on July 9, with the company realizing it had hacked Hugging Face after the breach was disclosed on July 16. Midjourney acquired the AI-powered astrology app Co-Star, with founder Banu Guler joining as chief design officer. Meta introduced agentic capabilities to Meta AI, enabling it to plan, research, create presentations and proactively complete tasks using connected apps.