Figma confirmed a partnership with Anthropic to launch Code to Canvas, a feature that converts code generated in AI tools into fully editable designs within Figma’s platform. Summary Anthropic released Claude Sonnet 4.6 in conjunction with the Figma partnership. Claude Sonnet 4.6 is being touted as Anthropic’s most capable Sonnet model yet. . Safety researchers described the model as having strong safety behaviors with no signs of major concerns around high-stakes forms of misalignment. The feature allows users who build interfaces using AI agents to import those interfaces directly into Figma’s canvas, according to the company’s announcement. Code to Canvas connects AI coding tools with Figma’s workflow, enabling teams to refine designs, compare options, and coordinate design decisions. Anthropic released Claude Sonnet 4.6 in conjunction with the partnership, stating the model demonstrates improved performance in computer use, coding, design, knowledge work tasks, and data processing. The company made Sonnet 4.6 the default model for Free and Pro users within the Claude chatbot and Claude Cowork productivity tool. Claude Sonnet 4.6 is being touted as Anthropic’s most capable Sonnet model yet. It also features a 1M token context window in beta, according to Anthropic. Developers with early access preferred Sonnet 4.6 over Sonnet 4.5 by a wide margin due to improved consistency and adherence to instructions. Early testers also frequently preferred it to Claude Opus 4.5, the company’s most advanced model released in November 2025. In Claude Code testing, users preferred Sonnet 4.6 over Sonnet 4.5 approximately 70% of the time, according to Anthropic. Users reported the newer model read context before modifying code and consolidated shared logic rather than duplicating it. Users preferred Sonnet 4.6 to Opus 4.5 about 59% of the time, rating it as less prone to overengineering and more consistent at following instructions. Testers reported fewer false success claims, reduced hallucinations, and more consistent completion of multi-step tasks. The model’s 1 million token context window can process entire codebases, lengthy contracts, or dozens of research papers in a single request, according to the company. In the Vending-Bench Arena evaluation, which tests business simulation performance, Sonnet 4.6 invested heavily in capacity during the first ten simulated months before pivoting to focus on profitability, a strategy that helped it finish ahead of competitors. Anthropic stated it conducted extensive safety evaluations and found Sonnet 4.6 to be as safe as or safer than other recent Claude models. Safety researchers described the model as having strong safety behaviors with no signs of major concerns around high-stakes forms of misalignment. The company reported Sonnet 4.6 showed improvements across benchmarks and approaches Opus-level intelligence at a lower price point. Early customers reported improvements in frontend code and financial analysis, according to Anthropic. Customers described visual outputs as more polished with improved layouts, animations, and design elements compared to previous Sonnet models, requiring fewer iterations to reach production-quality results.
