Anthropic’s AI model struggles to gain users amid cheaper competition

Anthropic built one of the most capable AI models on the planet. Turns out, “most capable” and “most used” are two very different things.

The company’s flagship Claude models are running into a wall of cheaper alternatives, particularly from Chinese competitors like DeepSeek, that deliver roughly comparable performance at prices that make Anthropic’s look like a luxury tax.

The price gap is staggering

Anthropic’s top-tier Fable 5 model runs between $10 and $50 per million tokens for input and output. DeepSeek’s equivalent offerings clock in at under $1 per million tokens. That’s not a modest discount. That’s a 10x to 50x difference in cost for what many users consider similar results.

AI startup Lindy offers a clean case study in what happens when companies do the math. The company entirely transitioned its usage to DeepSeek, driven purely by the economics.

This pattern is repeating across the industry. Enterprises are increasingly adopting “model routing” strategies, where they direct simple tasks to cheap models and reserve expensive ones for only the most complex operations. The net effect: less traffic and less revenue flowing to premium offerings like Anthropic’s flagship products.

Anthropic’s counter-move

Between June and August 2026, the company launched Claude Opus 5, a mid-tier model priced at $5 per million input tokens and $25 per million output tokens. Average prices paid for Anthropic’s models have dropped by nearly 25% since mid-July.

Claude reached roughly 245 million monthly active users by mid-2026, and Anthropic’s annualized revenue hit around $47 billion by May 2026. The company has even surpassed OpenAI in business spending share, capturing 41% of enterprise AI budgets as of May 2026.

Those figures come alongside slowing growth rates, as price sensitivity becomes the dominant force in enterprise AI procurement. High-margin API usage faces sustained pressure as buyers realize they have alternatives.

The commoditization problem

Chinese models currently lead on intelligence-per-dollar metrics. Anthropic’s market share on specific AI platforms has seen significant year-over-year declines, even as its absolute user count grows.

Open-weight models add another layer of competitive pressure. Companies can run these models on their own infrastructure, avoiding API costs entirely and gaining full control over their data. For privacy-conscious enterprises, that’s not just a cost advantage. It’s a compliance advantage.

What comes next

Both Anthropic and OpenAI are reportedly preparing for potential IPOs, which adds an interesting wrinkle to the pricing pressure story. Going public requires demonstrating sustainable revenue growth and healthy margins, two things that get harder when your competitors are undercutting you by an order of magnitude.

Anthropic’s response so far has been to segment its product line, offering premium models for complex tasks and discounted options for everything else. Enterprise buyers are already building their AI stacks with cost efficiency as the primary filter, routing tasks to the cheapest model that can handle them and only escalating to premium options when necessary. Anthropic’s $47 billion revenue run rate proves it can sell AI at scale. The open question is whether it can keep those margins intact as the floor drops out from under AI pricing.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.