OpenAI and Anthropic are cutting prices on their latest AI models as they compete for clients who are increasingly turning to cheaper alternatives from Chinese developers. The move follows rising AI costs that have pushed companies to limit usage and explore lower-cost options, helping firms like Moonshot and DeepSeek gain traction across Silicon Valley and Europe.

OpenAI announced an 80 percent price cut for GPT-5.6 Luna, calling it its fastest and most affordable model. Anthropic introduced Claude Opus 5, marketing it as offering frontier-level intelligence at half the price of its top-tier Fable 5. According to Silicon Data's token price index, prices for leading US lab models have dropped by nearly a quarter since mid-July. Tokens, the units of data processed by these models, are a common basis for customer billing.

This pricing shift marks a departure for US labs that have traditionally competed on performance with proprietary closed models. The rise of capable, freely downloadable open-source models from China has added downward pressure on prices. The cuts come as both OpenAI and Anthropic prepare for possible initial public offerings at trillion-dollar valuations, with investors looking for signs that massive AI spending will yield returns.

Corporate clients are feeling the squeeze as both labs move some enterprise customers from flat subscriptions to usage-based billing, where costs scale with computational use. In response, some businesses have set caps on AI usage or are testing cheaper alternatives. Companies including DoorDash and Airbnb have said they now use Chinese-made models to reduce expenses. This coincides with a wave of Chinese model releases that have closed the performance gap with US systems, fueling concerns that American developers could lose customers despite heavy investment in maintaining their lead.

Source: Ars Technica