OpenAI's GPT‑5.6 and rivals push AI token‑cost efficiency

OpenAI unveiled its latest generative‑AI model, GPT‑5.6, emphasizing that each token delivers more intelligence and stronger performance per dollar. The company introduced three tiered versions—Sol, Terra and Luna—claiming superior cost‑effectiveness compared with the previous GPT‑5.5.

Meta released its Muse Spark 1.1 model with a developer pricing of $1.25 for input and $4.25 for output per million tokens, roughly a quarter of OpenAI’s Sol tier cost. SpaceX AI’s Grok 4.5, announced a day earlier, touted token‑efficiency about twice that of competing models. The race for lower token costs stems from corporate customers' growing concern over hefty AI usage bills. Companies such as Uber have exhausted AI‑related budgets within months, prompting a shift toward more economical solutions, including the use of Chinese AI services and in‑house chip development. OpenAI CEO Sam Altman told CNBC that “this year is the first year AI costs have become a hot topic,” underscoring the industry’s pivot toward value‑driven competition.