Update. I have decided to take a break from Twitter to recharge a bit. See you tomorrow for more on OpenAI’s dreams to scale intelligence and bring more capable models to as many people as possible.
It’s going to be a fun and token efficient week. https://t.co/zlZhsQ8HRZ
— angel (@angelbrodin) July 30, 2026
UPDATE: It was a drastic price reduction for GPT-5.6 Luna, which is now 80% cheaper, and Terra, which is now 20% cheaper.
An interesting set of statements from a couple of OpenAI insiders on X may reveal the finer points of an upcoming product release today (Thursday).
The conversation starts with Tibo Sottiaux, the product lead for Codex who is also known for posting updates and limit resets for ChatGPT Work.
He says they are shipping today and that the focus of the week is to make intelligence too cheap to meter — which is precisely where the squeeze lies for enterprise adoption.
They are worrying about skyrocketing token costs when deploying across large organizations, and aren’t all that interested in cutting edge science features. They want ubiquitous AI for their workflows on the cheap, and lots of models have come out lately to cater to that demand.
Following up is Angel Brodin, listed on LinkedIn as a Strategic Success Engineer at OpenAI. She emphasizes token efficiency and scaling intelligence to bring capable models to more people.
This could lead to a rather obvious conclusion, considering where the market is and what it wants — and OpenAI is bringing large amounts of compute and inference optimizations to the table.
OpenAI normally releases their products at 10:00 at Pacific Daylight Time, which is 19:00 across most of Europe, and 01:00 on Singapore Standard Time, and, of note, there has been no information on this on OpenAI’s official account, but there might well be little later in the day. Theoretically, it could be just a price cut.
Read more: The X post. Discussion on r/Singularity.