Anthropic has implemented a strict pricing policy for enterprise clients: once a customer exhausts their contractually agreed-upon token quota, all discounts are immediately terminated. Clients must then either negotiate a new agreement or pay the full list price for subsequent usage.
In contrast, OpenAI offers a more flexible grace period for its enterprise customers. When a client hits their usage limit, OpenAI allows them to continue using discounted rates for the remainder of the current month plus one additional month to finalize a new contract.
Anthropic currently provides discounts of approximately 15% off list prices for large enterprise clients, a group that includes firms like Meta and Cursor. The company reports a gross margin of roughly 75% on its existing model sales business.
OpenAI is aggressively pursuing enterprise market share by reducing prices for models accessed through intermediaries like OpenRouter and Vercel by 50%. Recently, OpenAI CEO Sam Altman announced that the latest model generation features a 50% reduction in per-token pricing.
The competitive landscape has led both companies to include 'share of wallet' clauses in their contracts, requiring clients to allocate the majority of their AI budget to a single provider. However, consultants note that these clauses are difficult for the vendors to verify or enforce.
Despite OpenAI's aggressive pricing, Anthropic maintains a lead in revenue growth. Large enterprises such as Palantir and Nvidia continue to utilize both providers for automation and research tasks, despite implementing restrictions to protect their intellectual property.
