Summary: Recent pricing changes by major LLM providers like OpenAI and Anthropic highlight a shift toward more transparent and usage-based billing, impacting both individual users and enterprises.

The AI landscape is evolving rapidly, and with it, the cost structures for large language models (LLMs) are undergoing significant changes. Recently, whispers of profitability have started circulating around Anthropic, with rumors suggesting they may soon hit their first profitable quarter. Alongside this, a growing number of companies are finding themselves surprised by the rising costs of LLM usage—especially as more teams integrate these tools into their workflows.

This shift in pricing models is not just a financial concern; it reflects a deeper trend in the industry. Both OpenAI and Anthropic have clearly found product-market fit, leading to a surge in enterprise adoption. However, what’s becoming increasingly clear is that these models aren’t as cheap as many developers assumed—especially when used extensively.

Take a look at my own experience: I subscribe to the $100/month Max plan from Anthropic and the $100/month Pro plan from OpenAI. For someone who uses coding agents regularly, these plans offer great value. But when I ran the ccusagetool to estimate my API token costs over the past 30 days, the result was eye-opening: $2,180.16 worth of tokens for just $200—a fantastic deal, but one that only applies if you’re using these tools heavily.

What I didn’t realize was that enterprises often don’t get similar discounts. In fact, recent changes in pricing models suggest otherwise. Anthropic transitioned its Enterprise plan from a “Claude seats include enough usage for a typical workday” model in August 2025 to a $20/seat/month structure with additional API costs. This change, which occurred in November 2025, has caught many existing customers off guard during contract renewals.

OpenAI followed a similar path, updating its Codex pricing in April 2026 to align with API token usage rather than per-message pricing. While the new model offers more flexibility, it also introduces a more transparent—but potentially pricier—cost structure for heavy users.

As these pricing strategies evolve, developers and organizations must stay informed and reassess their LLM usage patterns. The era of low-cost experimentation may be ending, but with better transparency and structured plans, there’s still room for innovation.

Tags: #AI #LLM #Tech #Startup #Enterprise

Source: https://simonwillison.net/2026/May/27/product-market-fit/