SILICON VALLEY — Anthropic and OpenAI have each released artificial intelligence models that sharply cut prices while delivering top-tier performance. Analysts say the companies led with half-price models as corporate customers move to reduce spending on tokens, the unit by which AI usage is measured.
Anthropic on the 22nd released Claude Opus 5.5, a lower-cost model that it says performs at the level of Claude Fable 5.1, its flagship model.
Anthropic said it is the first model the company has introduced since it called for slowing the pace of frontier AI development, adding that the model matches Fable 5.1's performance on most tasks at 40% lower cost than its predecessor, Opus 5. On benchmarks for coding and knowledge work, it scored even higher than Fable 5.1.
OpenAI the same day unveiled GPT-6 Sol and GPT-6 Luna, models that sit below its flagship GPT-6 Astra. Application programming interface (API) pricing for the models is half that of their predecessors. OpenAI said that while Astra showcased a new generation of intelligence, Sol and Luna expanded the frontier of cost efficiency.
Companies that had competed to expand token usage earlier this year shifted strategy toward cost efficiency as AI budgets came under strain, and are increasingly turning to less expensive open-source and open-weight AI. That has prompted analysts to conclude that the two companies are leading with value-for-money models to defend their market dominance and to pull off successful initial public offerings.
Separately, according to The Information, Microsoft plans to discount subscriptions to Copilot, its AI agent for businesses, by 30% to 50%. The discounts take effect in October, with about 30% off for customers buying more than 1,000 licenses and up to 50% off for those buying more than 10,000.
