Anthropic launched Claude Opus 5.5, the first model in its new 5.5 family. It features a 1-million-token context window and costs $4 per million input tokens and $20 per million output tokens. The model underwent external evaluations by METR and Frontier Design.
Nvidia released Isaac ROS 5.0, a software stack for robotics developers. The update adds AI-agent skills for environment configuration and software migration, with support for ROS 2 Lyrical and the Jetson edge-computing platform.
Qualcomm unveiled the Snapdragon 8 Elite Gen 6 and Extreme Gen 6 processors, built on a 2-nanometer node. The Extreme variant supports on-device mixture-of-experts models exceeding 30 billion parameters. Motorola, Xiaomi, and ZTE are among the initial partners.
Palo Alto Networks introduced Unit 42 Continuous Frontier AI Defense, a service utilizing Anthropic and OpenAI models to scan corporate systems for vulnerabilities. Meanwhile, Microsoft disrupted the EvilTokens phishing platform, which compromised over 12,000 accounts.
OpenAI hired former Patreon executives, including co-founder Sam Yam, to lead a new Creator Product group. The team aims to develop tools for creators, with further details expected at the upcoming OpenAI DevDay on September 29.
The FBI is investigating a data breach involving the FBIJobs.gov portal, where hacking group ShinyHunters allegedly stole 2 terabytes of data. Additionally, Google released Chrome 154, addressing 108 security vulnerabilities, including 11 critical flaws.
