en.Wedoany.com Reported - On April 17 local time, Anthropic published a blog post officially announcing the launch of the visual creation tool Claude Design. Developed by Anthropic Labs, the tool is being gradually rolled out in a research preview to Claude Pro, Max, Team, and Enterprise subscribers, driven by the latest flagship vision model Claude Opus 4.7. In the blog post, Anthropic clearly stated that Claude Design targets users who "have ideas but no design background," helping entrepreneurs, product managers, marketers, and others quickly turn ideas into visual results, while also providing professional designers with broad exploration space.

The core interaction mode of Claude Design is conversation-driven creation. After users describe their needs in natural language, Claude generates an initial version. Subsequent adjustments can be made through further conversation, using embedded annotations to specify changes, or directly dragging sliders to control visual effects. Input methods cover text prompts, image uploads, and multi-format documents such as DOCX and PPTX. The platform also provides a web capture tool to directly extract design elements from existing websites to ensure prototypes align with actual products. Output formats support export to Canva, PDF, PPTX, or standalone HTML files. Completed designs can be handed off directly to Claude Code for the development process via a handover package.

Automatic application of brand systems is a key capability for team scenarios. During the initial onboarding phase, Claude Design can read enterprise code repositories and design files to automatically build a dedicated design system for the team. In subsequent projects, it automatically applies unified color schemes, fonts, and component specifications. Teams can maintain multiple design systems and continuously optimize them. Anthropic stated that more integrations with existing tools will be introduced in the coming weeks. The product is currently still in the research preview stage, and the final feature scope and commercialization terms of the official version have yet to be clarified.

Claude Opus 4.7 was released on April 16, just two months after the launch of Opus 4.6. According to Anthropic's official announcement, Opus 4.7 ranks first on the Vals Index comprehensive evaluation with a score of 71.4%, a significant leap from the previous best score of 67.7%. It also ranks first in multiple benchmark tests including Vibe Code Bench, Vals Multimodal, Finance Agent, and SWE-Bench Pro. On the software engineering benchmark SWE-bench Pro, Opus 4.7 improved from Opus 4.6's 53.4% to 64.3%, and on the standard programming task SWE-bench Verified, it improved from 80.8% to 87.6%. In terms of visual reasoning capabilities, Opus 4.7 jumped from 69.1% to 82.1% in the Vals Multimodal evaluation, supporting the parsing of images with a maximum long edge of 2576 pixels, approximately 3.75 million pixels, more than three times that of all previous Claude versions. Pricing remains consistent with Opus 4.6, at $5 per million tokens for input and $25 per million tokens for output. Opus 4.7 is now officially available across all Claude products and API interfaces, and has been integrated into Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry services.

Claude Design covers the entire application workflow from concept to delivery. According to internal demo cases, Anthropic employees used the tool to create a 3D interactive Earth visualization page, where 168 flight routes between 112 cities were connected by glowing arcs, while automatically generating six slider controls for users to adjust parameters like grid density, fill color, and rotation speed. In another demo, after a user uploaded an event record document, Claude Design organized it into a five-page landscape orientation welcome booklet containing a cover, itinerary, and packing list. In efficiency tests, Olivia Xu, a product designer at the online education platform Brilliant, stated that their team's most complex page required over 20 prompts in other tools to complete, but only 2 prompts in Claude Design. A product manager from the data monitoring platform Datadog reported that "the prototype was already runnable before the meeting ended."

The launch of Claude Design has attracted widespread attention in the design tool space. Anthropic told TechCrunch that Claude Design is positioned as complementary to tools like Canva, not a replacement, targeting user scenarios that require a quick transition from idea to visual result, rather than starting from professional design tools. Canva launched an MCP service for Claude in July 2025, allowing millions of users to directly create Canva designs within Claude conversations. Regarding Figma, in March 2026, it opened its Canvas Agent feature, allowing AI agents to directly edit Figma files and use components, variables, and annotations to create design resources. Additionally, Figma announced a partnership with OpenAI in February, integrating Codex into its design platform to enable AI-driven code generation. The Anthropic Labs team is co-led by Instagram co-founder Mike Krieger and is responsible for incubating and testing experimental AI products. Claude Design is another productivity-focused specialized product launched by this team following Claude Cowork.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com