Top-line takeaways and TSO verification:
Source 1 (Gizmodo, citing The Information): Anthropic is changing Claude Enterprise subscriptions from a “fixed fee of up to $200 per user per month” to “$20 per user per month plus compute/usage-based billing.”
Source 2 (CNBC): In old contracts, Anthropic’s standard/premium seats were “fixed monthly seats with built-in usage allowances,” and these seats have been marked as unavailable for new enterprise contracts; the new enterprise plan charges per seat and then bills token consumption at API rates.
Source 3 (VentureBeat): Claude Code has been defaulted to medium effort because users reported that it was consuming too many tokens; this change has been disclosed in the changelog and in the prompt shown when users open Claude Code.
TSO conclusion: The three sources agree on the core fact that Anthropic is tying enterprise Claude pricing more closely to token/compute consumption. However, on the question of whether Claude has been “weakened,” the available sources only support the presence of user feedback and localized default-behavior adjustments; they do not confirm any overall performance decline or the company’s underlying motive.
Shared confirmed facts:
Anthropic is changing enterprise contracts and billing for Claude Enterprise.
The old fixed monthly seat model with usage allowances is being phased out for new enterprise contracts.
The new plan directly links charges to token/API usage.
The market and users are already debating changes in Claude’s performance and token consumption.
Main differences or disagreements:
Different pricing wording:
Source 1 says pricing is shifting from “up to $200 per user per month” to “$20 per user per month plus compute usage.”
Source 2 describes the old model as standard/premium seats with “built-in usage allowances,” and the new model as “per seat + token consumption billed at API rates.”
These are directionally consistent, but the details are phrased differently.
The level of confirmation regarding “worse performance” or “nerfing”:
Source 3 only confirms that Claude Code’s default mode was changed to medium effort, explaining that this was in response to feedback about excessive token usage.
The provided sources do not confirm that the overall Claude model was intentionally downgraded, broadly “nerfed,” or that any performance change was directly caused by billing policy.
Motivation and scope of impact:
None of the three sources clearly states Anthropic’s full commercial motivation.
The provided sources do not allow confirmation of the impact across all enterprise customers, all product types, or all usage scenarios.
Background and analysis:
The core of Anthropic’s move is not simply a price increase. It is a shift in Claude Enterprise’s cost structure from a relatively fixed seat subscription to a hybrid model of “seat fee + usage-based billing.” As a result, enterprise users will no longer face only account-based costs; they will also see extra charges tied to token consumption. For companies that rely on high-frequency, long-context, or automated workflows, token costs will more directly reflect usage intensity, which can make actual operating costs feel higher.
Meanwhile, the controversy over whether Claude has become “worse” is, in the provided sources, better understood as a change in defaults and usage policy rather than proof of an across-the-board decline in model capability. Source 3 only says Claude Code was switched to medium effort because users complained it was consuming too many tokens, and that this was publicly disclosed. That means Anthropic is indeed adjusting the balance between resource consumption and output efficiency in at least this Claude Code scenario. But the sources do not support the conclusion that Claude’s overall quality has declined.
Three-source summary:
Source 1: Enterprise Claude subscriptions are shifting from fixed monthly fees to compute-usage-based billing.
Source 2: Old allowance-based seats are no longer available for new enterprise contracts; the new plan bills per seat plus token/API rates.
Source 3: Claude Code defaulted to medium effort in response to user complaints about excessive token consumption, and this has been publicly disclosed.
Conclusion:
Taken together, the three sources confirm that Anthropic is reshaping Claude Enterprise pricing and usage-based billing. But the debate over whether Claude has been “weakened” remains, for now, at the level of user feedback, localized default-setting changes, and public controversy. The provided sources do not offer enough evidence to confirm whether the model’s overall performance has declined, whether the change was driven by business strategy, or what the full impact on enterprise customers will be.