Key Highlights

  • OpenAI cut prices for its GPT-5.6 Luna model by 80%, while Anthropic launched Claude Opus 5 at half the price of its flagship Fable 5 model.
  • Average prices for leading US AI models have fallen nearly 25% since mid-July, according to Silicon Data's token price index.
  • Companies including DoorDash and Airbnb have said they have begun using Chinese-made AI models to reduce costs.
  • The price cuts come as OpenAI and Anthropic plan initial public offerings at valuations of $1 trillion or more.

US Labs Cut Prices on Mid-Tier Models to Compete

OpenAI and Anthropic have cut prices on their mid-tier AI models as rising costs push some Business customers toward cheaper alternatives from Chinese developers including Moonshot and DeepSeek. OpenAI cut the price of its GPT-5.6 Luna model by 80%, from $1 to $0.20 per million input tokens and from $6 to $1.20 per million output tokens. Anthropic launched its Claude Opus 5 model at $5 per million input tokens and $25 per million output tokens, half the price of its flagship Fable 5 model, and called off a planned September price increase for its Sonnet 5 model.

Average prices paid by customers for leading US models have fallen almost a quarter since mid-July, according to Silicon Data's token price index.

Enterprise Billing Shifts Add to Cost Pressure

The price cuts mark a shift for US AI developers, which had until now competed primarily on model performance rather than price. The change comes as some enterprise customers face rising costs from a shift away from flat subscription pricing toward usage-based billing tied to computational resource consumption. Companies including DoorDash and Airbnb have said they have started using Chinese-made models to help manage AI spending, as Chinese labs have narrowed the performance gap with leading US systems.

Comparisons Complicated by Differing Model Efficiency

Benchmarking firm Artificial Analysis found that Anthropic's Opus 5 model, run at "medium" effort settings, delivered similar performance and cost per task to Moonshot's Kimi K3 at "max" effort. Headline token prices do not provide a straightforward cost comparison, since more capable models can sometimes complete tasks using fewer tokens or fewer attempts. The price cuts come as OpenAI and Anthropic separately plan initial public offerings at valuations of $1 trillion or more, with investors seeking evidence that the industry's heavy AI spending can generate sustainable returns.