Anthropic Unveils Claude Sonnet 4.6 as Its Most Advanced AI Model Yet
AI company Anthropic, which recently made headlines when its workplace tool triggered a significant sell-off in IT stocks, has officially launched Claude Sonnet 4.6. In a detailed blog post, the creators of Claude announced that this new model represents a comprehensive upgrade across multiple domains, including coding, computer use, long-context reasoning, agent planning, knowledge work, and design.
Pricing and Availability Details
Claude Sonnet 4.6 is now the default model for Free and Pro plan users on claude.ai and Claude Cowork. The pricing structure remains unchanged from Sonnet 4.5, with a starting cost of $3 per million tokens for the Free plan and $15 per million tokens for the Pro plan. This ensures that users can access enhanced capabilities without any increase in expenditure.
Key Features and Performance Enhancements
The latest iteration brings substantial improvements in coding performance. Anthropic claims that tasks which previously required the more powerful Opus-class model, such as complex office and real-world work scenarios, can now be efficiently handled by Sonnet 4.6. This advancement makes high-level AI assistance more accessible and cost-effective for a broader range of users.
In terms of computer use capabilities, Sonnet 4.6 builds on Anthropic's general-purpose models that interact with software through clicking and typing, rather than relying solely on APIs. On the OSWorld benchmark, which evaluates AI performance in real-life applications like web browsers and code editors, Sonnet models have demonstrated consistent gains. Early users report near human-level proficiency in tasks such as navigating spreadsheets, completing multi-step web forms, and managing multiple browser tabs simultaneously.
Anthropic emphasizes that Sonnet 4.6 performs more consistently across various benchmarks and approaches Opus-level intelligence while maintaining affordability. Additionally, the model features an expanded context window, enabling it to process entire codebases, lengthy contracts, or extensive collections of research papers in a single request. This capability enhances long-term planning, as evidenced by strong results in the Vending-Bench Arena evaluation, where the model showcased effective decision-making over extended simulated timelines.
Safety and Accessibility
The company conducted extensive safety tests on Sonnet 4.6 and asserts that it is as safe as or safer than recent Claude models. This focus on safety ensures reliable and secure deployment across various applications.
Claude Sonnet 4.6 is available across all Claude plans, including Claude Cowork, Claude Code, the API, and all major cloud platforms. Notably, the free tier has been upgraded to default to Sonnet 4.6, now including features like file creation, connectors, skills, and compaction, making advanced AI tools more accessible to a wider audience.
