OpenAI cuts price of GPT-5.6 luna by 80% and GPT-5.6 Terra by 20%

• Says Terra and Luna usage now consumes fewer credits
• API pricing is USD 0.20/mln input tokens, USD 1.20/mln output tokens for Luna
• API pricing is USD 2/mln input tokens, USD 12/mln output tokens for Terra
• Sol pricing, ChatGPT and Codex subscription prices and quota budgets remain unchanged
• Adds that fast mode for GPT-5.6 Sol replaces priority processing in the API 

Steep API price cuts on flagship models have been a recurring feature of the AI buildout, and past episodes of this kind have tended to reflect a mix of genuine inference cost declines and competitive pressure on token pricing rather than demand weakness. The asymmetry in this move is notable: the cheaper tier cut far more aggressively than the premium one, with subscriptions and the top model left untouched, which reads as steering volume toward the low end rather than broad discounting. Historically the debate for the listed complex, Microsoft included, has split between those treating falling token prices as margin compression on AI monetisation and those treating it as a volume accelerant, and prior rounds have tended to resolve in favour of the demand elasticity read in the near term. The mechanism that matters is whether lower unit prices translate into faster enterprise API adoption, since that is what ultimately underwrites the capex cycle the hyperscalers are funding. Worth watching are follow-on pricing moves from rival labs, which in past cycles have come quickly, and any management commentary on utilisation versus realised price per token. The unchanged subscription pricing suggests consumer revenue is not the pressure point here.