OpenAI wants TV-level prices for ads inside ChatGPT

OpenAI is about to find out just how much a chatbot is really worth to advertisers, and the early signs point to a very expensive experiment. ChatGPT ads are rolling out in the coming weeks, and OpenAI is reportedly asking brands to pay TV-level prices to show up inside your conversations — without yet offering the kind of deep performance data that marketers are used to.

Here’s the basic shape of the deal: early reports say OpenAI is pitching a price of around $60 per 1,000 impressions for ads inside ChatGPT, a CPM that’s roughly three times what many advertisers pay on Meta’s platforms. That puts ChatGPT closer to live sports and premium streaming inventory than to typical social or search ads, at least on paper. For a media buyer, that’s a bold opening move from a brand‑new ad product that hasn’t yet proven its click‑through rates, conversion power, or long‑term return on investment.

At the same time, OpenAI is making a very deliberate promise to users: your chats come first, ads come second. In a recent policy post, the company laid out a set of principles for advertising, including “answer independence,” which means ads are not supposed to influence what ChatGPT actually tells you. Sponsored content will sit in a clearly labeled section at the bottom of responses, rather than being woven into the answer itself, and advertisers won’t get to steer the model’s outputs behind the scenes.

Related /

OpenAI is also trying to position this as a way to keep AI cheap — or at least cheaper — for most people. Ads will only appear (at least at first) for adults in the U.S. using the free tier or the lower‑cost ChatGPT Go plan, while Plus, Pro, Business, and Enterprise subscribers stay ad‑free. The pitch is straightforward: if you don’t want to pay for a subscription, the presence of ads helps cover the massive compute bill that comes with running a popular AI model at scale.

The trade‑off is on the data side. Unlike Google or Meta, which live and die on granular campaign reporting, OpenAI is starting with a pretty bare‑bones analytics package. Early advertisers are expected to see high‑level stats like total impressions and total clicks, but not what happens after that — no detailed conversion tracking, no clear attribution to purchases or sign‑ups. That’s partly because OpenAI is trying to back up a core promise: it says it will never sell your ChatGPT conversations to advertisers and will keep that data private, which dramatically limits how far it can go with the usual ad‑tech tracking tricks.

From a marketer’s perspective, that makes ChatGPT feel more like a premium awareness channel than a performance‑marketing workhorse, at least in its first iteration. You’re paying a lot to show up in a high‑intent, highly trusted space, but you’re not yet getting the dashboard full of conversion funnels and lookalike audiences that you’d expect from a mature ad platform. That’s a big ask at $60 CPM, especially when most new ad products launch with discounts to entice brands to experiment rather than premiums that place them above established channels.

OpenAI clearly thinks the context justifies the price. When people are asking ChatGPT for help choosing a laptop, planning a vacation, or deciding which online course to take, they are sitting in what advertisers call a high‑intent moment — they are actively trying to make a decision. OpenAI’s early tests focus on shopping‑style ads that show sponsored products or services relevant to the current conversation, displayed in a separate, clearly marked box under the model’s answer. If this works as envisioned, brands won’t just be interrupting a feed; they’ll be stepping into the moment someone says, “What should I buy?” and getting a chance to respond in context.

That kind of “conversational ad” is very different from a static banner on a website or a skippable pre‑roll on a video. OpenAI has already hinted at a future where you can tap on an ad and keep talking — ask follow‑up questions, compare options, narrow down features — all inside the same interface. In theory, that blends the intent of search, the personal tone of chat, and the targeting ambitions of social into a new format that traditional platforms don’t really offer yet.

But this is also where things get tricky. If the model is answering your question and an advertiser is trying to sell you something in the same window, how do you make sure users can tell where neutral advice ends and commercial messaging begins? OpenAI’s answer so far is labeling: the company says ads will be visually distinct, will not alter the underlying answer, and will be subject to policies that block them from appearing around sensitive topics like mental health or politics and for users under 18. It’s a cautious rollout, and it has to be, because any hint that ChatGPT’s core answers are pay‑to‑play would damage the trust that made the product popular in the first place.

Behind all of this is a bigger strategic question: how does OpenAI diversify its revenue so it isn’t just a subscription SaaS company or an enterprise vendor? The company already sells higher‑tier plans and enterprise access, and those will remain ad‑free, but ads give it another lever that’s more familiar to the broader internet economy. If ChatGPT can become a high‑margin ad surface in its own right, OpenAI gets a path to fund more compute and model development without pushing subscription prices ever higher.

Still, this is not a guaranteed win. Advertisers are famously pragmatic: if the performance isn’t there, premium pricing doesn’t last long. Brands will want to see real evidence that showing up in ChatGPT conversations moves the needle on awareness, intent, or sales, and they’ll push for more sophisticated reporting and targeting over time. And regulators are already examining how AI systems use data, so any drift toward more invasive tracking could land OpenAI in the same privacy debates that have dogged the rest of the ad industry.

Related /

Lawmakers are worried ChatGPT ads could cross a line

For regular users, the immediate experience may feel closer to what you already see on search engines than to the noisy ad feeds of social platforms. You’ll get an answer, and then, sometimes, a small sponsored block below that tries to capitalize on whatever you just asked. The real test will be whether those placements feel helpful — “oh, that’s actually a useful product link” — or intrusive, like a brand is eavesdropping on a conversation you thought was just between you and an AI.

Right now, OpenAI is walking a tightrope: charging like a premium ad network, acting like a privacy‑conscious consumer product, and trying to invent a new format on the fly. Over the next few months, as those first campaigns go live on the free and Go tiers in the U.S., we’ll see who blinks first — the marketers being asked to pay top dollar, or the AI company betting that the future of advertising lives inside a chat window.