AI labs raise prices without raising prices (playcode.io)

Recent developments within AI labs have sparked controversy as companies like OpenAI and Anthropic have effectively raised their prices while maintaining the same token rates, leading to higher costs for users. OpenAI’s release of GPT Image 2.6, for instance, demonstrated a substantial increase in output tokens—3.3 times more for the same prompt compared to its predecessor, GPT Image 2.5. Anthropic has similarly adjusted its tokenizer, resulting in about 30% more tokens produced for the same input. This strategy allows companies to enhance their profit margins without overtly increasing prices, which could lead to user dissatisfaction once the cost discrepancies are realized. The implications of this trend are significant for the AI/ML community, raising concerns about transparency and user trust. As developers and researchers rely on these models for various applications, they must now navigate not only the performance differences but also the hidden cost implications of the token inflation. With models like Claude Fable 5.2 offering cheaper per-token costs while utilizing significantly more output tokens, users will need to closely monitor their expenses as they make decisions about which models to adopt for their tasks. This shift in pricing strategy reflects a growing sophistication in AI economic models but also necessitates vigilance from users to avoid unexpected increases in operational costs.