Subscribe to Our Newsletter

Success! Now Check Your Email

To complete Subscribe, click the confirmation link in your inbox. If it doesn’t arrive within 3 minutes, check your spam folder.

Ok, Thanks

Anthropic launches Claude Design

Anthropic released Claude Design, a conversational tool that converts prompts into interactive prototypes and visual assets for paid Claude subscribers.

Defused News Writer profile image
by Defused News Writer
Anthropic launches Claude Design

Anthropic launched Claude Design, making the tool available in research preview to paid Claude subscribers and aiming to turn text prompts into polished visual work including designs, interactive prototypes, slide decks, and marketing collateral.

The product is built on Claude Opus 4.7 and is part of Anthropic’s push from foundation models into full applications.

The company said in an announcement it is gradually rolling access to Pro, Max, Team, and Enterprise subscribers, and that Claude Design links to other Anthropic products, including Claude Code, to move projects from prototype to production code.

Claude Design generates an initial layout from a natural-language brief and refines it through chat, inline comments, text edits, and custom adjustment sliders. During onboarding the tool reads a team’s codebase and design files to build a reusable design system.

Export options include internal URLs, folders, Canva, PDF, PPTX, and standalone HTML files, while a web-capture tool pulls elements from live sites. Anthropic cites early customers (including Brilliant and Datadog) that reduced multi-step design cycles into fewer prompts and single conversations.

Opus 4.7 improves vision and coding performance and is deliberately less capable than Anthropic’s more restricted Mythos-class model. Anthropic stated in its release that it "experimented with efforts to differentially reduce" the new model's cyber capabilities during training and ships Opus 4.7 with safeguards that detect and block high-risk cybersecurity requests.

The model posts a 64.3% on SWE-bench Pro and shows a 13% improvement on Anthropic’s 93-task coding benchmark; it accepts images up to 2,576 pixels on the long edge and scored 98.5% on a partner visual-acuity test versus 54.5% for the prior model.

For enterprises, Anthropic says it stores the generated design-system representation rather than source files, does not upload linked local code, and will not train on customer data.

Claude Design is included at no extra cost with Pro, Max, Team, and Enterprise plans; Opus 4.7 API pricing remains $5 per million input tokens and $25 per million output tokens. Anthropic notes collaboration and editing still have rough edges and will use user feedback to guide wider availability.

The recap

  • Anthropic launched Claude Design in research preview for paid subscribers.
  • Opus 4.7 accepts images up to 2,576 pixels on the long edge.
  • Anthropic is in early IPO talks with major investment banks.
Defused News Writer profile image
by Defused News Writer

Explore stories