OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why? (minimaxir.com)

OpenAI and Anthropic have significantly reduced the prices of their large language models (LLMs), with some cuts reaching up to 80% in just a few months. OpenAI's GPT-5.6 Luna, once priced at $1, now stands at $0.20, while Anthropic’s Claude Sonnet 5 has also become a permanent low-price offering. The motivation behind these drastic price reductions appears to stem from increased market competition and the emergence of cheaper, effective open-weight models. As LLMs become more commodified, both companies are facing pressure to maintain their user bases and competitive edge by making their models more accessible. This pricing shift is particularly noteworthy for the AI/ML community as it could fundamentally alter the cost dynamics of using LLMs for various applications. High cache reading costs have traditionally been a barrier, but Anthropic's recent models have drastically reduced these, leading to total usage costs dropping by about 70-75%. As companies gear up for IPOs, the approach of lowering prices despite potential immediate revenue implications suggests a calculated strategy to attract users and usage volume, while also preparing for increased scrutiny from investors. Ultimately, these developments signal a rapidly evolving landscape in the LLM market where affordability and performance are becoming increasingly critical.