OpenAI Turns Token Pricing Into a Weapon
OpenAI’s dramatic GPT-5.6 pricing cuts are a deliberate move to turn raw model power into a low-cost utility, using cheaper token pricing to lock in developers, squeeze rivals, and shift the AI market from experimental hype to high-volume, everyday use.
The headline move: OpenAI slashed GPT-5.6 Luna API costs by 80%, dropping input tokens from USD 1 (approx. RM4.60) to USD 0.20 (approx. RM0.92) per million and output tokens from USD 6 (approx. RM27.60) to USD 1.20 (approx. RM5.52) per million. Terra followed with a 20% cut to USD 2 (approx. RM9.20) per million input tokens and USD 12 (approx. RM55.20) per million output tokens, down from USD 2.50 (approx. RM11.50) and USD 15 (approx. RM69.00). These are not cosmetic tweaks; they are margin sacrifices aimed at volume and dominance. When a company slashes API prices three weeks after launch, it signals urgency, not comfort.
A Calculated Strike in an Emerging AI Price War
This is not a goodwill discount; it is a pre-emptive strike in an AI price war. One outlet framed it bluntly: “Let the AI price wars begin,” noting that OpenAI is cutting token prices by as much as 80% on GPT-5.6 Luna to “starve out” rival labs. At the same time, Open-source models, including several from China, have been putting pressure on the market by offering lower-cost alternatives for businesses and developers.
The timing is telling. The price changes arrived as enterprises become more cautious about AI spending and rivals intensify competition. Token pricing is now a competitive front in its own right, not a billing detail. Instead of fighting only on benchmark scores, OpenAI is signaling that it will also fight on OpenAI API costs, using its scale and infrastructure efficiency to make it painful for smaller or less capitalized players to match these rates.
Chinese AI Competition and the New Economics of Tokens
The move is especially pointed because it lands amid rising Chinese AI competition. One report notes that OpenAI has fired “a veritable volley across the bow” of a growing number of labs, targeting the “budding open-weight AI economy” with these cuts. At the same time, some rival token pricing is moving upward rather than down; Moonshot’s Kimi K3 is priced at USD 3 (approx. RM13.80) per million input tokens and USD 15 (approx. RM69.00) per million output tokens.
This contrast matters. When OpenAI can offer Luna at USD 0.20 (approx. RM0.92) per million input tokens while some competitors sell input at USD 3 (approx. RM13.80), it reframes expectations for what a high-end model should cost. Token pricing becomes a strategic lever: squeeze margins now to consolidate demand, then depend on scale and efficiency to survive the fallout. Nobody should assume rivals will quietly accept this; but for the moment, OpenAI has dragged the reference price for premium AI sharply downward.
One Billion Users, Two Million Businesses – and a Margin Squeeze
OpenAI did not announce these GPT-5.6 pricing cuts in a vacuum. The company disclosed that its models now reach more than one billion active users and more than two million businesses, and it tied that milestone directly to a push to make AI more affordable through lower prices and better infrastructure efficiency. In its own words, “When the cost of useful intelligence falls, more work becomes worth doing”.
On the ground, the impact is immediate: Luna and Terra now consume fewer credits in paid Codex and ChatGPT Work subscriptions, while subscription prices and quota budgets stay unchanged. Lower OpenAI API costs encourage developers to run bigger contexts, more agents, and more experimental workloads. Yet this volume play is financially risky. One analysis notes that OpenAI generated USD 13.07 billion (approx. RM60.12 billion) in revenue but lost USD 21 billion (approx. RM96.60 billion) in 2025, with only around 50 million of 900 million weekly users paying for a subscription. Cheaper tokens widen adoption but also deepen the profitability challenge.
From Capability Race to Cost and Access Race
The deeper shift is that AI competition is no longer only about who has the most capable model; it is about who can offer the most useful intelligence per dollar, at scale. OpenAI itself links price cuts to a flywheel: better intelligence drives broader adoption, broader adoption supports more investment, and more investment improves intelligence and efficiency. Open-source AI models, including those from China, have accelerated this shift by proving that “good enough” models at low cost can win real workloads.
For businesses and developers, the message is clear: the AI price war is now structural, not temporary. Lower token pricing makes more AI projects economically viable, from always-on agents to AI-heavy products. But it will also force tough choices among providers who cannot afford a race to the bottom. OpenAI’s GPT-5.6 pricing cuts mark a turning point where access and cost efficiency matter as much as benchmarks—and the winners will be those who can scale cheaply without collapsing under their own compute bills.
