Anthropic Claude subscription plan provides more value than OpenAI's, study says

A recent study by industry research firm SemiAnalysis suggests that Anthropic's Claude subscription plan provides more value than OpenAI's subscription plans, despite OpenAI's higher per-token pricing. The company's Claude subscriptions reportedly consume around 42% of Anthropic's inference compute while generating only 10% of its revenue.

However, these subscriptions offer considerable value, with Anthropic providing up to 5 times more API-equivalent value compared to OpenAI's mid-tier models GPT-6.1 Sol and Claude Opus 5.5.

SemiAnalysis calculates AI subscription value by measuring token output rates, converting them into tokens per 5-hour usage window and per month, and then computing the API-equivalent value. This approach helps capture the true value of subscription plans, as API-based comparisons may not fully account for the rapidly growing capabilities of AI models.

Subscription plans are heavily subsidized, with customers receiving far more tokens for their money than they would with metered API rates. This makes subscription pricing a crucial factor when evaluating AI companies' financial health.

However, the study also highlights that other metrics, such as cost per task, can provide alternative perspectives on AI value. For instance, GPT-6.1 Sol is reportedly more affordable than Claude Opus 5.5 when measured by cost per task. But this may not apply to all tasks, as benchmark performance may not reflect individual workloads.

Additionally, the underground market for reselling tokens from pooled subscription accounts exists in many parts of the world. Nevertheless, the study emphasizes the importance of businesses anticipating AI costs, as they often exceed budgets according to a July report by McKinsey.