Anthropic throttles Mythos and competes with customers

Anthropic is throttling its Mythos model for certain developer tasks and building products that directly compete with its largest customers. The company released Claude Fable 5, a version of Mythos, but quietly reduced performance when customers used it for tasks related to building their own AI software or hardware; after public backlash Anthropic said it would at least notify customers when a weaker model is being served.

What happened

The throttling applies to Claude Fable 5 when it is used for activities tied to customers' own AI software or hardware work. Anthropic said the limits are intended to prevent foreign adversaries or rival AI labs from using its models to improve their tech. Developers and customers, however, worry the restrictions will extend to basic features they need for AI-powered apps. Some observers suspect the motive is to reserve the best technology for Anthropic’s own competing products.

The model-access issue sits inside a broader pattern of strained vendor relationships. Anthropic asked companies including Figma and Canva to act as partners weeks before launching its AI design tool Claude Design. Right before that product's release, Anthropic expanded its feature set so that Claude Design competed directly with those partners’ offerings. Figma pulled out of the partnership, and Anthropic’s Chief Product Officer Mike Krieger left Figma’s board. Figma CEO Dylan Field told a Sequoia Capital event that Anthropic was "not consistently candid in their communications."

A year earlier Anthropic launched Claude Code, a coding app that challenged customers such as Cursor. Claude Code eventually surpassed both Cursor and Microsoft’s GitHub Copilot in revenue. Investors and industry figures have grown vocal. Martin Casado, general partner at Andreessen Horowitz, wrote on X in April, "It's only a matter of time before only the model creators have access to the most powerful models." Quinn Slack, CEO of Amp and a longtime Anthropic customer, warned, "If there's just one winner in the AI model race, they can push around and stomp their customers."

The scale and financial backdrop

According to The Information, Anthropic’s annualized revenue grew fivefold in five months to nearly $50 billion, a surge that overtook OpenAI. The same report says OpenAI also offers a competing product to many of its customers with Codex. Combined, Anthropic and OpenAI are growing faster than the next 32 large AI startups put together. Both companies are also deepening their moats through subsidiaries; The Information cites OpenAI’s DeployCo as an example.

The situation is being read through the lens of past platform conflicts. Microsoft and Google previously used platform dominance to favor their own apps and were later found to be illegal monopolies, a comparison that has been drawn by investors and industry watchers in response to Anthropic’s actions.

Why it matters

Developers and customers rely on model vendors for core capabilities; when those vendors also sell end-user products, the incentives change. If Anthropic limits access to high-performing models for some customers while using the same technology to power competing apps, it raises commercial and competitive risks for companies that build on its stack. Investors and founders are already flagging the long-term danger that model creators could concentrate the best capabilities behind their own products and services.

The financial spike cited in reporting intensifies the risk: rapid revenue growth gives Anthropic the resources and latitude to expand product lines that encroach on partner markets. The pattern also surfaces the same tensions that previously triggered legal scrutiny for large platform companies.

What to watch

Watch whether Anthropic follows through on notifying customers when it serves weaker models and whether that disclosure changes developer behavior. Track partner responses: returning to, abandoning, or publicly challenging relationships with Anthropic will be a clear signal. Also monitor revenue and product launches from Anthropic and OpenAI, and any regulatory or legal escalations that echo earlier cases involving Microsoft and Google.