The most powerful mid-tier model to date, Claude Sonnet 4.6, has been formally released by Anthropic. It offers a thorough improvement in coding, computer use, long-context reasoning, agent planning, knowledge work, and design, all at the same cost as its predecessor. With API prices remaining at $3/$15 per million input/output tokens, the model is now the default on Claude Cowork's and Claude.ai's Free and Pro plans.

Interestingly, Sonnet 4.6 comes with a 1 million token context window in beta, which is big enough to accommodate dozens of research papers, lengthy legal contracts, or entire codebases in a single request.

Enhancement of Coding Performance According to developer testing, users cited improved context comprehension and less code duplication as reasons for choosing Sonnet 4.6 over Sonnet 4.5 about 70% of the time in Claude Code sessions. Significantly less prone to overengineering, fewer hallucinations, and more consistent follow-through on multi-step tasks were the main reasons why users chose Sonnet 4.6 59% of the time over Opus 4.5 Anthropic's frontier model from November 2025. Sonnet 4.6 has the smallest performance difference between the Sonnet and Opus tiers in any Claude generation, scoring 79.6% on the SWE-bench Verified benchmark, up from 77.2% on Sonnet 4.5 and nearly 80.8% on Opus 4.6.

Customers also mentioned that there were notable improvements in frontend code and financial analysis, and that the visual outputs were noticeably more polished, had better layouts, animations, and design sensibility, and required fewer iterations to reach production quality. Computer Use Takes a Dramatic Leap Sonnet 4.6 nearly matches Opus 4.6 at 72.7% and nearly doubles GPT-5.2's 38.2% score on OSWorld-Verified Anthropic's benchmark for autonomous computer use across real software, such as Chrome, LibreOffice, and VS Code, with a score of 72.5%. With a roughly 5× improvement over sixteen months, this represents a startling increase from 14.9% when computer use was first introduced in October 2024.

Early users report human-level capability in tasks such as navigating complex spreadsheets and completing multi-step web forms across multiple browser tabs.

With Sonnet 4.6 performing similarly to Opus 4.6 on safety evaluations, Anthropic has also strengthened the model's resistance to prompt injection attacks, a major security concern in agentic workflows. Platform Availability and Updates Sonnet 4.6 in beta on the Claude Developer Platform facilitates context compaction, extended thinking, and adaptive thinking by automatically summarizing older context as discussions get closer to their limits. With the use of dynamic code execution, web search and fetch tools can now automatically filter and process results, enhancing token efficiency and response quality.

Available on Pro, Max, Team, and Enterprise plans, the Claude in Excel add-in now supports MCP connectors for integration with platforms such as S&P Global, LSEG, PitchBook, Moody's, and FactSet.

All Claude plans, Claude Code, the API, and popular cloud platforms like Microsoft Azure AI Foundry now support Claude Sonnet 4.6. For daily cybersecurity updates, developers can use the model string claude-sonnet-4-6, LinkedIn, and X to access it through the API. To have your stories featured, get in touch with us.