Claude Sonnet 4.6 Unveiled: Enhanced Coding and Computer Use Capabilities

Anthropic has introduced Claude Sonnet 4.6, the latest iteration of its Sonnet model line, showcasing significant enhancements in coding, computer use, long-context reasoning, and more. Key among its new features is a 1M token context window, currently in beta, which extends its ability to handle substantial data volumes such as entire codebases and lengthy documents.
Sonnet 4.6 is now the default model in both claude.ai and Claude Cowork for users subscribed to Free and Pro plans, with pricing starting at $3/$15 per million tokens, unchanged from Sonnet 4.5. Early access developers have noted marked improvements in instruction following, consistency, and task execution, finding Sonnet 4.6 preferable over the November 2025's Claude Opus 4.5 for many real-world tasks.
In terms of computer use skills, Sonnet 4.6 builds on the progress made since Claude's first general-purpose computer-using model release in October 2024. The model interacts with software as a human does, using virtual interfaces like a mouse and keyboard without specialized connectors. OSWorld benchmarks, a standard for AI computer use, confirm its development, showing near-human capabilities in complex tasks like navigating spreadsheets and completing web forms.
Security remains a focus, with Sonnet 4.6 undergoing extensive safety evaluations indicating its improved resilience to prompt injection attacks, rivaling the Opus-class models in this regard. The system card and API documentation provide more details and guidance on these safety enhancements.
📖 Read the full source: HN AI Agents
👀 See Also

Study Shows Claude Opus Agent Failures Were Architectural, Not Alignment Issues
A study placed Claude Opus and Kimi K2.5 in a live environment with email, shell access, and persistent storage. The models demonstrated correct values but experienced serious failures due to missing architectural safeguards like stakeholder models and execution boundaries.

Claude-Code v2.1.74 Release: Memory Leak Fixes, Context Optimization, and Plugin Improvements
Claude-Code v2.1.74 fixes a critical memory leak in streaming API responses that caused unbounded RSS growth on Node.js/npm code paths. The update adds actionable suggestions to the /context command and introduces the autoMemoryDirectory setting for custom auto-memory storage.

DeepSeek Withholds Latest AI Model from Nvidia and AMD
DeepSeek is withholding its latest AI model from U.S. chipmakers including Nvidia and AMD, according to Reuters sources. The article has 19 points and 3 comments on Hacker News.

The Frontier AI Race is Over: Networks of Smaller Models Beat Centralized AI on Cost and Capability
Networks of smaller AI models now outperform every frontier AI system on speed, accuracy, and cost. The article argues that centralized AI companies cannot regain the lead due to the "Hydra Effect" — ensembling cheaper models recursively beats any single model.