OpenAI launched GPT-6 Astra in early September 2026 as its flagship large language model with standard API pricing of $10 per million input tokens and $50 per million output tokens. The output rate has held steady amid strong benchmark results on computer use, coding, and reasoning tasks, yet traders anticipate a cut before year-end due to competition from OpenAI’s own lower-priced GPT-6.1 Sol at $2/$10 and rival frontier models. Historical precedent shows gradual reductions on prior flagships, while recent safety-related cancellation of a higher-tier 6.1 Astra upgrade and expanded availability of cheaper variants reinforce expectations of downward pressure. Key upcoming catalysts include earnings reports, developer conferences, or new model releases that could signal pricing shifts.