Data scientist Max Woolf highlights a sweeping trend of major artificial intelligence labs cutting foundational model token prices by up to 80% over recent months. The analysis points to fierce competition from open-weights alternatives such as DeepSeek and upcoming models from Moonshot AI as a primary downward driver on enterprise pricing. Significant reductions in cache read costs have drastically lowered the effective cost of running autonomous coding agents and multi-turn prompt chains. While industry observers debate whether these aggressive cuts are loss-leading or driven by genuine hardware and inference efficiency gains, consumer spend has shifted. The shifting market dynamics arrive as industry leaders like Anthropic and OpenAI eye landmark public market debuts, raising questions about long-term profit margins.

Full analysis : Why are OpenAI’s and Anthropic fighting a LLM token price war.