The new CC, an AI agent built for families
Google has developed a new AI agent called CC designed specifically for family use. The AI aims to assist with household tasks and provide family-oriented support. (blog.google)
Claude Opus 5.5
anthropic.com1246 points836 comments
Anthropic’s Claude Opus 5.5 is a new high-capability model in the Claude 5.5 family that improves performance, safety, and cost-efficiency over Opus 5. It matches or edges Claude Fable 5.1 on many real-world tasks and scored best yet on Anthropic’s automated behavioral audit, with external evaluation from Frontier Design and METR. Opus 5.5 is more resistant to prompt injection, less prone to taking hard-to-reverse actions, and uses the same tightened safeguards deployed on other capable models; vetted organizations can apply for controlled access for life‑sciences and cybersecurity work. The model communicates more naturally and prioritizes key information, which testers found made long sessions easier to follow. Runtime is more than 30% faster than Opus 5, and typical costs are 40% lower: input tokens $4/million, output $20/million, cache reads $0.20/million (with a fast mode available at higher rates).
Substantively, Opus 5.5 shines on agentic coding, computer use, and knowledge-work benchmarks while using fewer tokens per task. Concrete examples include completing a 680,000‑line code migration in under a day, auditing a 200,000‑line codebase in under three hours (vs. 20+ hours for Opus 5), and translating HAProxy from C to Rust in 9.5 hours at 51% lower cost than a comparator. Benchmarks cited include Terminal‑Bench (66.4%), FrontierCode (54.4%), CursorBench (57.8%) and strong GDPVAL knowledge scores; Anthropic highlights that efficiency gains drive much of the practical advantage. Claude Sonnet 5.5 and Haiku 5.5 are slated to follow.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
More in AI
AI staff complain of mental toll over fears of threat to society
AI staff have reported experiencing mental health challenges due to concerns about the potential societal threats posed by artificial intelligence. These fears relate to the broader impact of AI development on society and human well-being. (ft.com)
The current balance of power in open models
Open language models include open-weight and open-source types, with Chinese companies leading in open-weight models since 2025. The competition between American and Chinese models influences the global AI landscape and its commercial viability. (interconnects.ai)
No Sloptober
No Sloptober challenges individuals to abstain from using LLM-based tools throughout October to develop personal skills and awareness of AI's limitations. Participants are encouraged to reflect on their reliance on AI, improve their coding abilities, and reassess the value of automation in their work and learning processes. (no-sloptober.com)
Open-Weight AI Models Seize Token Lead, but Proprietary Still Make the Money
Open-weight AI models now process 56% of tokens in production, up from less than 10% in December 2025. However, proprietary systems still generate most of the revenue, as they cost significantly more per inference. (techstrong.ai)
LLM Ass Bench
LLM ASSBENCH is a platform that evaluates large language models (LLMs). It currently shows no matches or data for prompts entered. (assbench.com)