OpenAI cuts prices for two of its GPT-5.6 AI models as companies grow sensitive to costs
CNBC ● Covered by 14 sources
OpenAI just cut prices on two GPT-5.6 models, only three weeks after launch. Companies are getting stingy with AI budgets, and rivals are racing to look cheap.
Based on reporting by CNBC — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Three weeks. That's how long OpenAI let its new GPT-5.6 Terra and Luna models sit at their launch prices before slashing them. Terra drops 20%, now $2 per million input tokens and $12 per million output tokens. Luna gets hit harder, an 80% cut down to 20 cents per million input tokens and $1.20 per million output tokens. Sol, the flagship of the three, keeps its price untouched.
The timing tells you everything. Enterprises that once threw money at AI without blinking are now asking harder questions about return on investment, and OpenAI knows it. The company's own line is that each new generation should do more work for less money, which sounds like a mission statement but reads more like a defensive move.
That defensiveness makes sense given who's circling. Moonshot AI, a Chinese startup, released an open-weight model called Kimi K3 this month that beat leading American models on some benchmarks, which got Silicon Valley's attention fast. Anthropic answered with Claude Opus 5, priced at half of its own Claude Fable 5 despite performing similarly on coding and knowledge tasks. Microsoft's Satya Nadella spent part of Wednesday's earnings call talking up his company's cheap models, right after shipping a low-cost cybersecurity model. Google, meanwhile, rolled out three new models this month, including Gemini 3.6 Flash, which it claims undercuts Kimi K3 and other Chinese offerings on cost per task.
The bigger story here is the end of what people started calling tokenmaxxing, the phase after ChatGPT's 2022 debut when companies pushed employees to use AI constantly and worry about the bill later. Some of those bills reached into the billions, and now the reckoning has arrived. Every major lab is suddenly competing on price as much as on raw capability, and that's a very different fight than the one everyone expected a year ago.
My take — AI-written commentary, not fact-checked reporting
Nobody in this industry wants to admit the free-spending honeymoon is over, but three price cuts in one month from three different labs says it plainly enough. The real winner isn't whoever tops a benchmark chart, it's whoever survives the moment enterprises start reading their invoices closely. Open-weight competition from China forcing this kind of discipline on Silicon Valley is the sort of pressure the industry needed and mostly avoided until now.
Read more about this at: CNBC
