Anthropic is preparing to release Claude Opus 4.7 and a new AI tool for generating websites and presentations, according to a report from The Information published Tuesday. Both products could ship as early as this week.
The model is the less interesting half of this announcement. Opus 4.7 follows Opus 4.6, which Anthropic released in early February, which followed Opus 4.5 last November. At this cadence, a new Opus every three to four months, the version bump barely qualifies as news. What caught my attention was the design tool, and what it did to stock prices within hours of the report landing.
The market reaction
Figma fell 6%. Wix dropped nearly 5%. Adobe slid 2.7%. GoDaddy lost 3%. All on the back of a single-source report about a product nobody has seen yet. That is either a market that's been bracing for this kind of move for months, or one that panics at shadows. Probably both.
The tool, per The Information's source, would let technical and non-technical users create presentations, websites, landing pages, and other products using natural language prompts. A direct competitor to Gamma, the AI presentation startup valued at $2.1 billion after its November 2025 Series B, and Google Stitch, which Google Labs launched at I/O 2025 and recently expanded into an AI-native design canvas.
But Anthropic is not Gamma or Google Labs. Anthropic is a company whose core product is the model itself. And that distinction, a foundation model company building first-party productivity tools on top of its own AI, is what rattled investors.
Why this is different from the model upgrade
Opus 4.7 is an incremental step. We know roughly what to expect. Opus 4.6 hit 65.4% on Terminal-Bench 2.0 and 72.7% on OSWorld, according to Anthropic's product page. It expanded the context window to one million tokens in beta and proved capable of writing a C compiler in Rust from scratch (a $20,000 experiment by Anthropic researcher Nicholas Carlini, per Wikipedia's entry on Claude). Opus 4.7 will presumably push those numbers higher.
The design tool is a strategy shift. Anthropic has been, until now, a model company that sells API access and a chat interface. Claude Code expanded into developer tooling. The PowerPoint integration that came with Opus 4.6 hinted at productivity ambitions. But a standalone tool for creating websites and presentations? That is Anthropic competing directly with its own customers and partners.
Gamma, remember, uses Claude as one of its 20+ AI models. If Anthropic ships a competing product, what does that relationship look like in six months?
The leaks told us this was coming
None of this should be surprising if you have been paying attention to Anthropic's recent, uh, information security record. On March 31, source maps were accidentally included in Claude Code's npm package, exposing roughly 500,000 lines of internal TypeScript. The leak, first reported by Fortune, revealed version strings for Opus 4.7 and Sonnet 4.8, along with references to a next-generation model family called Mythos and a new tier above Opus called Capybara.
Days earlier, a misconfigured CMS had made nearly 3,000 unpublished documents publicly searchable. Among them: a draft blog post describing Capybara as "larger and more intelligent than our Opus models." Security researchers Roy Paz of LayerX Security and Alexandre Pauwels of the University of Cambridge found the exposed data. Anthropic called it "human error."
Two leaks in one week. Both attributed to mistakes. Both revealing major product plans.
Anthropic confirmed it is testing Mythos, calling it "a step change and the most capable we've built to date." But Opus 4.7 and Mythos/Capybara appear to be different things entirely. Opus 4.7 is the next incremental update within the existing Claude 4.x family. Capybara would be a new fourth tier, sitting above Opus, at a price point one analyst blog estimated could reach $10-15 per million input tokens (current Opus runs $5).
What the design tool actually means for the market
Gamma has 70 million users and $100 million in ARR. It hit those numbers with 52 employees. Google Stitch, meanwhile, is still free during its Labs phase, running on Gemini 2.5 Pro. The AI presentation and website space is already crowded: Beautiful.ai, Tome, Canva's AI features, Microsoft Copilot in PowerPoint.
Anthropic entering this market makes a different kind of noise. This is a company with over $64 billion in total capital raised. Annualized revenue reportedly hit $19 billion as of March 2026. And it is widely expected to pursue an IPO, possibly as early as October, at a valuation in the range of $400-500 billion.
Launching a consumer-facing design product before an IPO makes financial sense. It broadens the revenue story beyond API calls and subscriptions. It gives the S-1 a line item that reads "productivity tools" instead of just "model access." Wall Street likes product diversification.
For Figma, the threat is less about Anthropic stealing their users today and more about the signal it sends. If foundation model companies start building vertical tools on their own models, the "AI wrapper" critique that has haunted startups for two years starts applying to bigger companies too. Figma's $12.5 billion in market cap took a visible hit on what amounts to a rumor.
What I do not know
The Information's report comes from a single unnamed source. No screenshots. No product demo. No confirmed name for the design tool. We don't know if it is a standalone app, a feature within claude.ai, or something else. We don't know pricing. We don't know if "as soon as this week" means Wednesday or if it means Anthropic is still deciding.
Anthropic has not commented.
The company's next confirmed public benchmark is whatever Opus 4.7 scores on the usual gauntlet: SWE-bench, Terminal-Bench, OSWorld. Those numbers will matter more for developers than any design tool. But the design tool, if it ships, will matter more for Anthropic's IPO narrative.
And that, I suspect, is the point.
