OpenAI Cuts GPT-5.6 Luna Pricing 80% as Model Price Competition Intensifies

OpenAI cut GPT-5.6 Luna pricing by 80%, to $0.20 per million input tokens and $1.20 per million output tokens, alongside a 20% cut to Terra pricing (now $2/$12) and a 2.5x-faster Fast mode for Sol — announced July 30, 2026, the same week OpenAI reported crossing 1 billion active users.

Why the Timing and Scale of the Cut Matter

An 80% price cut on a flagship-tier model is a significant move, not a routine adjustment — it's a direct response to intensifying competition from both open-weight Chinese models (DeepSeek, Kimi K3, GLM) undercutting on cost, and rival proprietary labs like xAI and Anthropic competing on capability-per-dollar. For any business running high-volume AI workloads on Luna specifically, this is a direct, immediate cost reduction with no migration required.

• Luna's new pricing puts it firmly in the cost tier competing with smaller, cheaper open-weight models — while still being a proprietary, managed API
• The pairing with the 1-billion-user milestone suggests OpenAI is optimizing for volume and lock-in at this price point, not maximizing per-token margin
• Enterprise AI budgets that modeled Luna at its prior pricing should revisit those forecasts — an 80% swing is large enough to materially change a build-vs-buy or model-selection decision made even a few months ago