OpenAI Says Models Now Reach 1B+ Users After GPT-5.6 Cuts

OpenAI told the Wall Street Journal that its models now reach more than one billion active users and more than two million businesses, an announcement it paired with fresh price cuts on the smaller tiers of its GPT-5.6 family. Luna, the fastest and most affordable tier, dropped 80% to $0.20 per million input tokens and $1.20 per million output tokens, from $1 and $6. Terra, the mid-tier, dropped 20% to $2 in and $12 out per million tokens, from $2.50 and $15. Sol, the flagship, was left alone.

The shape of that move is more interesting than the raw numbers. The cuts land exactly on the segment enterprises now scrutinise most, high-volume token consumption where cost per call decides whether a workflow ships or dies in review. Per Quartz's summary, OpenAI attributed the reductions to efficiency gains from GPT-5.6's own internal development, including the model's role in optimizing production software and improving speculative decoding, which reduced end-to-end serving costs by 20% and increased token-generation efficiency by more than 15%.

Why this matters if you are budgeting for AI usage right now: cheaper Chinese open-weight models from the likes of Moonshot, Alibaba and DeepSeek have been eating into US enterprise token share, and OpenAI is defending the middle of its price ladder rather than the top. Leaving Sol untouched is itself a signal, that OpenAI still believes the frontier tier can hold its price while the volume tiers absorb the war.

The honest caveat is that a headline user count is a soft number. What the reporting does not give you is a split between paid and free users, whether the two million businesses figure counts paid API customers or ChatGPT team seats, or anything about retention. Take the growth claim as the company's framing, not settled unit economics.

For builders, the immediate upside is straightforward. High-volume workflows that were uneconomic on Luna at $1/$6 are back on the table at $0.20/$1.20, and that is quietly where the next wave of shipped AI features gets built.