GPT-5.5 costs 49 to 92 percent more than its predecessor, depending on the input length

Editor's take

OpenAI's pricing adjustments for GPT-5.5, initially presented as a cost optimization strategy tied to response length, have demonstrably led to significant expense increases for users. OpenRouter's analysis reveals that actual costs have climbed between 49% and 92% for many users, directly challenging OpenAI's stated efficiencies.

This price hike is a critical development for developers and businesses relying on OpenAI's models. It signals a potential shift in the economics of LLM deployment, forcing organizations to re-evaluate their budgets and potentially seek more cost-effective alternatives like Anthropic's Claude models, especially for applications with longer input requirements. The opaque nature of model pricing and performance claims is being exposed, highlighting the need for greater transparency in the LLM market.

The immediate challenge for OpenAI is to reconcile its pricing claims with observed user costs, potentially impacting adoption rates if current trends persist. Future developments to monitor include whether OpenAI revises its pricing structure, offers tiered performance/cost options, or if competitors like Anthropic capitalize on this perceived shift in value by offering more predictable and demonstrably cheaper alternatives.