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's next flagship model leaks, and it's bringing a design tool with it

Claude Opus 4.7 could drop this week, alongside a product that rattled shares of Figma, Adobe and Wix

Ian Lyall profile image
by Ian Lyall
Anthropic's next flagship model leaks, and it's bringing a design tool with it
Photo by Phil Hearing / Unsplash

Anthropic is preparing to release its next flagship AI model, Claude Opus 4.7, along with a new design tool that would mark the company's first move into visual productivity software. Both products could ship as early as this week. Anthropic has not confirmed or denied the plans.

The breadcrumbs were already there

The report did not land in a vacuum. On March 31, Anthropic accidentally left source maps in Claude Code's npm package, exposing roughly 500,000 lines of internal TypeScript. The code spread across the internet within minutes, downloaded and mirrored before the company could contain it.

Buried in those files were references to Opus 4.7 and Sonnet 4.8, along with a next-generation model family called Mythos and a new tier above Opus called Capybara. Codenames, including Fenick for the Opus series and Capra for Sonnet, were also identified.

Days earlier, a misconfigured CMS had made nearly 3,000 unpublished Anthropic documents publicly searchable, among them a draft blog post describing Capybara as "larger and more intelligent than our Opus models." Two leaks in one week. Both are attributed to human error.

A model upgrade, but the bigger story is the design tool

Opus 4.7 is an incremental step within the existing Claude 4.x family. Its predecessor, Opus 4.6, scored 65.4% on Terminal-Bench 2.0 and 72.7% on OSWorld, expanded the context window to one million tokens in beta, and proved capable of writing a C compiler in Rust from scratch. Opus 4.7 will presumably push those numbers higher.

The design tool is the more significant signal. It would let both technical and non-technical users create presentations, websites and landing pages using natural language, putting Anthropic in direct competition with Figma, Adobe, Wix and startups like Gamma.

Until now, Anthropic has been a model company that sells API access and a chat interface, with Claude Code as its only real product extension. The design tool would represent a step into consumer-facing productivity software, a very different kind of business.

Markets moved before anything launched

Shares of Figma, Adobe, GoDaddy and Wix each dropped between 2% and 4% after the plans became public. Polymarket reset to 98% probability of a launch by the end of June.

The market reaction is telling. Anthropic has raised capital at a valuation approaching $800 billion. When a company with that kind of firepower enters your market with a tool that lets anyone describe a website and have it built, incumbents tend to notice.

Opus 4.7 is not the top of the range

One detail worth noting is that Opus 4.7 is not Anthropic's most capable model. That distinction belongs to Claude Mythos, which is currently being tested by select partners for security vulnerability research. Anthropic maintains a dual-track strategy, with Opus 4.7 set for broad public release while Mythos remains under controlled access.

The pace of releases is striking. Anthropic has shipped three major model updates in three months, Opus 4.5 in February, 4.6 in March, 4.7 in April, each arriving alongside new product features and enterprise integrations.

What to watch this week

If Opus 4.7 and the design tool land as reported, Anthropic will have moved from pure model provider to something closer to a platform company in the space of a single quarter. That is the strategic shift that matters here, not the benchmark scores.

The model upgrade was expected. The ecosystem play was not.

Ian Lyall profile image
by Ian Lyall