OpenAI said on Thursday it will cut prices for two GPT-5.6 models after
improving system efficiency: Luna will be reduced 80% to 0.20 dlr per MLN input
tokens and 1.20 dlr per MLN output tokens; Terra will be reduced 20% to 2 dlr
per MLN input tokens and 12 dlr per MLN output tokens. The cuts are reflected in
Codex and ChatGPT Work billing; the top-tier GPT-5.6 Sol will not be discounted
but OpenAI said it will offer a faster API option for Sol. OpenAI and Anthropic
have both adjusted pricing, rate limits and other usage policies as
next‑generation inference models drive sharply higher token consumption in
longer-running agent tasks.
