OpenAI Cuts GPT-5.6 Prices as Rivals Close In

Three weeks after launching the GPT-5.6 family, OpenAI has cut the price of its cheapest model by 80 percent. On July 30 the company dropped GPT-5.6 Luna from 1 dollar and 6 dollars per million input and output tokens to 20 cents and 1 dollar and 20 cents. Terra, the middle tier, came down about 20 percent. Sol, the flagship, held steady at 5 and 30 dollars.

The timing tells the story. A lab does not carve 80 percent off a model it shipped three weeks earlier unless it feels the floor moving. Luna is pitched at roughly 85 percent of Sol's quality for a fraction of the cost, and that is exactly the band where the pressure is fiercest, because it is the band where cheap open weight models have been landing all summer.

Fast mode for the flagship

Alongside the cuts, OpenAI introduced a Fast mode for Sol on the API, replacing what used to be called Priority Processing. It runs the top model at up to 2.5 times the normal speed for twice the price, aimed at developers who need low latency and will pay for it. VentureBeat read the whole package as a shift in where the fight now lives, away from raw capability and toward cost per token.

Where the pressure comes from

It is not hard to find the source. DeepSeek's V4 Flash and Alibaba's Qwen 3.8 Max, both covered here in the past two weeks, have been undercutting the American labs on price while narrowing the gap on quality. When a capable model is available cheaply and with open weights, a premium API has to justify itself on something other than being the only option. For buyers, this is the good part of a boom. The price of a token keeps falling.

The margin question

The open question is what these cuts do to OpenAI's own economics. The company is spending heavily on compute and has been candid that its frontier tiers are not cheap to serve. Cutting Luna to 20 cents may win volume, but volume at a thin margin is a strange comfort when your funding rounds are the largest in the industry's history. The price war is plainly good news for customers. Whether it is good business is a separate question.