Particle.news
Download on the App Store

OpenAI Cuts Prices on Two GPT‑5.6 Models

Efficiency gains in GPT‑5.6 let OpenAI lower mid- and low-tier costs while increasing how many tokens paid subscribers can use before hitting limits.

Overview

  • OpenAI on Thursday, July 30, 2026 cut prices for GPT‑5.6 Terra by 20% and Luna by 80%, bringing Terra to $2 per million input tokens and $12 per million output tokens and Luna to $0.20 per million input tokens and $1.20 per million output tokens.
  • The flagship Sol model kept its original price so OpenAI can preserve a premium tier for the highest‑performance uses.
  • OpenAI said efficiency gains in GPT‑5.6, including faster code and internal optimization, made the cuts possible and enable lower per‑token costs for many tasks.
  • The company also adjusted how Terra and Luna token usage counts against paid ChatGPT Work and Codex subscriptions so paid users can run more tasks before hitting subscription limits.
  • Industry observers say lower per‑token prices may boost adoption but could squeeze vendor margins and raise enterprise bills because firms are shifting to usage‑based pricing and agentic workflows that make cost per completed task harder to predict.