AI
Anthropic launches Claude Opus 5.5 as OpenAI ships GPT-6 Sol and Luna
Anthropic and OpenAI trade same-day model launches, with Opus 5.5 undercutting prices and GPT-6 Sol and Luna targeting cheaper tiers.
This edition was produced with artificial intelligence. Text and voice are generated automatically.
Anthropic launches Claude Opus 5.5 at lower prices and claims benchmark lead
Anthropic launched Claude Opus 5.5, the first model in a new family, saying it matches Claude Fable 5.1 on most tasks at significantly lower cost and runs faster than Opus 5. Anthropic benchmarks place it ahead of Fable 5.1 and OpenAI’s GPT-6 Astra on most tasks, including Terminal-Bench 4.0 at 66.4 percent and Humanity’s Last Exam at 67.7 percent with tools. Pricing falls to $4 per million input tokens and $20 per million output tokens, a 20 percent cut, with cache reads down 60 percent to $0.20. Anthropic says total operating costs should be about 40 percent lower than Opus 5’s. It is available on all major platforms, including AWS, Google Cloud, and Microsoft Azure.
OpenAI releases GPT-6 Sol and GPT-6 Luna with halved token prices
OpenAI launched GPT-6 Sol and GPT-6 Luna, two models following GPT-6 Astra and built with similar methods for gains in professional work, factuality, coding, computer use, and alignment. GPT-6 Sol targets everyday work at lower cost than Astra and makes about half as many mistakes as GPT-5.6 Sol, while GPT-6 Luna is cheaper still and optimized for fast, high-volume responses. Token prices fall versus 5.6 counterparts: Sol costs $2 per million input tokens and $10 per million output tokens, down by half, and Luna costs $0.10 and $0.50. Both roll out in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users, with Luna also available to Free and Go users in the desktop app. Neither is yet available in Chat mode.
OpenAI’s New GPT-6 Sol and Luna Models Bring Astra Improvements to Cheaper Tiers →
Meta patches Muse zero-day that let attackers control the AI agent
Meta patched a zero-day vulnerability in its Muse macOS app that could let an attacker take control of the AI agent. Security researcher Patrick Wardle found the bug, which used an undocumented Muse setting to redirect transcription processing from Meta’s servers to an attacker’s endpoint, giving access to the Muse account. Wardle’s proof-of-concept attacks let him take pictures and write malicious files to disk via Muse, often without alerting the user. Meta patched within hours of the Ars Technica report and said real-world risk was low because the exploit required local access. David Singleton of Meta Superintelligence Labs called it a local privilege escalation attack and said a hotfix was issued.
Meta patches Muse exploit that let attackers control the AI agent →
Amazon blocks Meta’s Muse agent from shopping on Amazon.com
Amazon blocked Meta’s Muse AI agent from Amazon.com after Meta rejected a request to remove the bot, with the block beginning Sunday night. Users trying to buy via Muse were told continued access by an unauthorized AI agent violates Amazon’s Conditions of Use. Amazon spokesperson Lara Hendrickson said third-party apps that purchase on behalf of other companies’ customers should operate transparently and respect service providers’ decisions. Amazon said it was not told in advance that Muse would access its store, that Muse does not identify itself when browsing, and that it appears to capture and store customer credentials and scrape account data. Meta said Muse has no visibility into passwords or payment methods and stores credentials securely. Muse launched September 8 and reached No. 1 among free U.S. App Store apps within a week.
Anthropic builds biology lab where Claude guides robots through drug experiments
Anthropic is building a biology lab in the San Francisco area to run physical drug experiments, moving AI-driven drug development beyond computer simulations. Claude will guide robots through experiments with minimal human involvement, though human oversight remains required for safety. Anthropic has shipped two tools for this: Claude Science, an AI workspace for researchers, and the Model Hardware Standard for controlling lab equipment. Eric Kauderer-Abrams, who runs Anthropic’s life sciences division, sees the biggest opportunity in diseases considered undruggable because no treatment exists, and says AI could speed up design of complex antibodies that hit multiple targets at once. The lab builds on Anthropic’s April acquisition of Coefficient Bio for about $400 million and the addition of Novartis CEO Vas Narasimhan to its board. Anthropic will not run its own clinical trials for now.
Anthropic is setting up a biology lab where Claude guides robots through drug experiments →