OpenAI just made two of its newest AI models a lot cheaper, and the timing tells you something about where the whole industry is headed.
About three weeks after releasing GPT-5.6 Terra and GPT-5.6 Luna to the public, the company knocked 20% off Terra and a whopping 80% off Luna. Terra's input tokens are now priced at $2 for one million, with output tokens at $12 for one million. Luna runs at 20 cents per one million input tokens and $1.20 for each million output tokens.
These are not tiny adjustments. An 80% cut in less than a month is a loud signal. OpenAI is trying to keep customers from walking away.
Get the market news that matters in a five-minute read with Market Briefs, our free daily newsletter
The rapid price reductions underscore a fundamental shift in the AI landscape: companies are no longer willing to pay premium prices without guaranteed returns. This trend has been accelerated by the rise of open‑weight models from Chinese startups like Moonshot AI, which offer comparable performance at a fraction of the cost. OpenAI's cuts are a strategic response to prevent customer churn and maintain market share against rivals like Anthropic and Google, who have also slashed prices in recent weeks.
OpenAI is being pushed to appeal to budget‑conscious clients, as businesses grow reluctant to implement pricey AI systems without a clear view of what they'll get back. Meanwhile, the company is also trying to hold its own against rivals from China and major tech players like Google and Microsoft, all of whom are promoting affordable AI options.
OpenAI ignited the artificial intelligence surge when it released ChatGPT in 2022, sparking a wave of U.S. companies scrambling to implement the tech and encourage employees to adopt it. This led to the "tokenmaxxing" period, in which bosses urged workers to maximize AI usage without fretting over expenses.
Yet as AI costs have exploded - occasionally reaching billions of dollars - numerous firms have been striving to curb their outlays. This change in behavior has not escaped the notice of AI creators, particularly given how fast Chinese open‑weight models have closed the gap with top‑tier proprietary systems.
Open‑weight AI models - which users can freely download, tweak, and host on their own systems - offer a budget‑friendly alternative for businesses. Earlier this month, Chinese startup Moonshot AI unveiled Kimi K3, an open‑weight model that beats top American products on certain benchmarks, triggering a rapid response from Silicon Valley. In the wake of that move, OpenAI's main rival Anthropic released Claude Opus 5, calling it their finest and most economical option for many tasks.
Priced at half the cost of Claude Fable 5 - which Anthropic debuted in June - Opus 5 delivers similar performance in coding and knowledge‑based assignments. Microsoft CEO Satya Nadella, during his company's Wednesday earnings call, repeatedly underscored the affordability of Microsoft's models, following the earlier launch of a cybersecurity model that the firm touted as both cheap and effective. Google, meanwhile, rolled out three new models this month with the explicit goal of beating rivals on price.
According to Google, its strongest new model, Gemini 3.6 Flash, is cheaper than Kimi K3 and other Chinese alternatives for each individual task.
Join Market Briefs, our free daily newsletter, for a quick daily rundown of the markets
