ChatGPT Ad Formats vs Other AI Assistants 2026

ChatGPT Ad Formats vs Other AI Assistants: A 2026 Buyer's Map

As of April 2026, ChatGPT ships sponsored shopping cards and conversational banners (live since February 9, 2026, $60 CPM, formerly $200K minimum, now $50K via the self-serve ads manager). Perplexity paused new advertisers in late 2025 after launching sponsored follow-up questions in November 2024. Gemini's AI Mode carries sponsored shopping placements; the core Gemini app remains ad-free per Google's VP of Global Ads. Copilot runs Performance Max placements in conversations with reported 73% higher CTR than traditional search. Claude is pledged ad-free per Anthropic's 2026 Super Bowl campaign. Independent networks like Thrad assemble equivalent prompt-triggered inventory across surfaces without the waitlists.

Case Study ->
Case Study ->
Case Study ->
Case Study ->
Case Study ->
Case Study ->

Case Study ->

ChatGPT Ad Formats vs Other AI Assistants 2026 | Thrad

Every major AI assistant now carries — or is about to carry — paid inventory. ChatGPT, Perplexity, Gemini's AI Mode, and Copilot each ship a different ad format with different access rules, pricing, and measurement maturity. Claude remains the deliberate holdout. This is the complete April 2026 map, honestly graded by what's available today vs waitlisted, with a look at how independent networks give buyers one way to reach the whole surface set right now.

Date Published

Apr 19, 2026

Date Modified

Apr 19, 2026

Start monetizing your AI app in under an hour

With Thrad, publishers go from first API call to live ads in less than 60 minutes. With fewer than 10 lines of code required, Thrad makes it easy to unlock revenue from your conversational traffic the same day.

Monetize your AI

In April 2026, there is no such thing as a single "AI assistant ad product." There are five surfaces — ChatGPT, Perplexity, Gemini, Copilot, and Claude — and each has made a distinct policy choice about ads, inventory, access, and labeling. This piece is the complete buyer's map: every shipped and in-pilot format, how each is priced and sold, what creative each accepts, and where independent networks like Thrad close the gap when a buyer needs one plan that covers the whole surface set today. Where OpenAI is gated behind a formerly $200K and now $50K minimum via its pilot ads manager, and where Perplexity has paused new advertisers entirely, the practical question for most brands is not "should I buy ChatGPT ads" — it's "which assistant inventory can I actually reach right now, and at what price."

What are the AI assistant ad formats in April 2026?

There are six distinct ad formats live or in confirmed pilot across the major AI assistants in April 2026: ChatGPT sponsored shopping cards, ChatGPT conversational banners, Perplexity sponsored follow-up questions, Gemini AI Mode sponsored shopping, Copilot Performance Max conversational placements, and independent-network prompt-triggered ads. Claude ships no ads by deliberate policy. Each format differs on access, minimum spend, labeling, and measurement.

The honest summary a buyer needs up front: only ChatGPT, Gemini AI Mode, Copilot, and independent networks are open to new advertisers right now. Perplexity is closed. Claude will not sell. And the two most frequently hyped formats — ChatGPT's self-serve ads manager and Gemini-app ads — are either very new (ChatGPT ads manager pilot) or not actually shipping in the form advertisers have been briefed on (Gemini app).

ChatGPT ad formats: what shipped February 9, 2026

OpenAI turned on ads inside ChatGPT on February 9, 2026 — not earlier despite years of speculation. The company's public post on advertising framed the decision as a way to "expand access to ChatGPT," tying free-tier sustainability directly to an ad-supported layer. Two formats are confirmed in-market, and a third (a self-serve ads manager) is in pilot.

1. Sponsored shopping product carousel

This is the flagship format. When a user asks ChatGPT a shopping, food, or product-recommendation question, the assistant can render a sponsored product card beneath its answer. The card shows brand logo, a "Sponsored" label, product name, price, stock status, and estimated delivery or prep time. Checkout is integrated with Etsy and Shopify for participating retailers, which is how ChatGPT collapses what used to be three funnel steps (search → landing page → cart) into one in-surface event.

Pricing: approximately $60 CPM headline rate, with actual clearing rates ranging from $18 to $65 depending on topic cluster and advertiser category, per ALM Corp's published pricing breakdown. That $60 figure has held consistent since the pilot opened in February 2026, placing ChatGPT inventory roughly in line with premium streaming and NFL broadcast CPMs — well above Meta's typical rates.

2. Conversational banner

The second format is more ambitious: a conversational banner below the answer where users can ask ChatGPT follow-up questions about the advertised product itself. In other words, the ad isn't just a static card — the assistant will field "does this come in blue?" or "is it good for kids?" style queries, with the retailer's product data grounding the reply. This is closer to a generative sales associate than a display unit, and it is the 2026 format most clearly native to an AI assistant.

3. ChatGPT ads manager (pilot)

In parallel with the launch, OpenAI opened a self-serve ads manager to a small pilot group. Per ppc.land's coverage, the minimum commit dropped from as high as $250K to $50K at the pilot open — still a non-trivial floor for most SMBs, and several multiples of what a new Google Ads advertiser can spend. The original $200K threshold was confirmed in an Adweek exclusive during the early beta.

Audience reach is capped by OpenAI's own subscription gating: ads appear only to Free and Go plan ($8/month) users who are logged in, 18+, and located in the US. Plus, Pro, Business, Enterprise, and Education users do not see ads. This is a meaningful constraint — it means ChatGPT ads reach a demographically different slice of the assistant's userbase than the paid subscriber base where most buying-committee members for B2B products actually sit.

ChatGPT advertising crossed $100 million in annualized revenue within six weeks of formally launching on February 9, 2026, at roughly $60 CPM served only to Free and Go-tier users in the US.

How does the Perplexity ad product actually work?

Perplexity's ad format is the oldest in the set — sponsored follow-up questions launched in November 2024 — and paradoxically the least accessible today. The format is conceptually elegant: when a user gets an answer on Perplexity, a set of AI-generated "People also ask"-style follow-up prompts appear beneath the answer, and the first slot can be sponsored. Importantly, advertisers do not write the question's phrasing; Perplexity's model generates it, and the AI still writes the answer when the user clicks. The advertiser pays on a CPM basis for impression share on topically matched follow-ups, per Perplexity's own announcement post.

In late 2025, eleven months after launch, Perplexity paused accepting new advertisers — the company said publicly that it wanted to reassess its ad ambitions. Existing early partners continue to run, but the self-serve ad platform that marketers expected has not shipped. For a 2026 buyer, this means Perplexity is effectively a closed surface — not because the company has no inventory, but because new-advertiser intake is currently gated to zero.

That gating is significant because Perplexity was, up to this point, the most open and transparent AI-assistant ad surface. Marketers who had built internal Perplexity test plans need a substitute, and the practical substitutes are ChatGPT (gated by $50K minimum and tier-cap), Copilot (requires Microsoft Advertising infrastructure), or an independent prompt-triggered network that can reach Perplexity-like inventory without a direct Perplexity contract.

Gemini ad formats: the AI Mode / Gemini app split

Google's position is the most confusing in the set, which is why it needs to be broken in half. There are two distinct Google surfaces:

Gemini app — the standalone chatbot. As of April 2026, the Gemini
app is ad-free, per Google's VP of Global Ads Dan Taylor in December
2025 and DeepMind CEO Demis Hassabis at Davos in January 2026. Google
briefed advertising clients on a planned 2026 rollout (per Adweek),
then publicly corrected the report via Search Engine Land: "There are
no ads in the Gemini app and there are no current plans to change
that."
AI Mode — Google Search's generative answer experience, launched
in March 2025 and expanded through 2026. AI Mode does carry sponsored
shopping placements, clearly labeled, and is bought through Google Ads
like every other Search product.

The distinction matters for media plans. An advertiser who "buys Gemini" is really buying AI Mode, which is a feature of Search — not the standalone Gemini chatbot where consumer assistant-style engagement concentrates. Google's SVP of Knowledge and Information, Nick Fox, publicly said Google is "not ruling out" Gemini-app ads, and the learnings from AI Mode are the leading input for whatever the company eventually ships on Gemini. But "eventually" is doing heavy lifting — the timeline is explicitly not now.

Copilot ad formats: the quietest maturation

Microsoft Copilot is the AI-assistant ad surface that has attracted the least 2026 hype and quietly become the most operationally mature. Performance Max campaigns from Microsoft Advertising automatically qualify for Copilot placements, with ads appearing directly beneath Copilot's AI-generated responses and labeled "Microsoft Advertising" + "Sponsored," per Microsoft's own product page.

Microsoft's reported performance data is notable. A November 2024 to May 2025 study said Copilot conversational placements generate 73% higher click-through rates and 16% stronger conversion rates than traditional search advertising, with customer journeys 33% shorter. The caveat — this is a vendor-published number, not a third-party audited study — but the numbers have held through subsequent 2026 coverage from MediaPost and fit directionally with what independent prompt-intent data shows.

2026 additions include a double-row sponsored carousel paired with organic cards for specific SKU surfacing, and interactive product exploration inside conversations — a direct parallel to ChatGPT's conversational banner format, shipping earlier and integrated with Microsoft Advertising's pre-existing ad sales infrastructure.

Why does Claude ship no ads in 2026?

Claude is the deliberate holdout. Anthropic ran a 2026 Super Bowl campaign positioning Claude as ad-free, with fictional sketches of chatbot conversations disrupted by inappropriate ads — a satirical jab at the OpenAI launch days earlier. Anthropic's public statement: Claude will remain free of sponsored links or third-party placements influencing responses.

The revenue side justifies the stance. Per CNBC's February 2026 reporting, Anthropic's 2025 revenue was $4.5 billion, predominantly from API access to coding startups (Cursor, Cognition) and enterprise customers like Microsoft and Canva. Claude Code and Cowork together represent at least $1 billion of that. With an API + enterprise revenue model that scales at Anthropic's growth rate, an ad layer adds little and costs trust. The decision is consistent.

For a buyer, this means Claude is simply not in the media plan. Not even as an "eventually" — Anthropic has been unusually direct. Any plan that claims Claude inventory access is either misrepresenting reality or, more commonly, routing Claude users toward web-publisher placements that Claude surfaces as citations. That's a valid adjacent path but it is not Claude advertising.

How do independent ad networks fit across these surfaces?

An independent AI-assistant ad network fills three buyer problems that platform operators can't or won't solve: cross-surface inventory, transparent pricing, and access without six-figure minimums. A network like Thrad aggregates prompt-triggered inventory across AI assistant surfaces and the web-publisher fringe that feeds their citation layer, then sells it as a single bookable plan with open CPM reporting.

Thrad's practical advantage as of April 2026: available today (no waitlist), cross-surface by design, transparent pricing (no $50K or $200K minimum), and an open measurement layer that reports impressions, clicks, and citation lift rather than proprietary "brand lift" metrics that only the platform sees. Where OpenAI's own product is gated and opaque, and Perplexity has closed new-advertiser intake, independent networks are the only path for a brand that wants AI-assistant reach on a regular media-plan timeline.

The full format comparison table

Surface	Format(s) live	Access for new advertisers	Minimum spend	Labeling	Who sees the ad
ChatGPT	Shopping cards, conversational banner	Open via Ads Manager pilot	$50K (was $200K)	"Sponsored" tag	Free + Go tier, logged-in US adults
ChatGPT ads manager	Self-serve pilot	Invite-only pilot	$50K	"Sponsored" tag	Same as above
Perplexity	Sponsored follow-up questions	Paused to new advertisers	N/A currently	"Sponsored" tag	US users, all tiers
Gemini AI Mode	Sponsored shopping	Open via Google Ads	Standard Google Ads floors	"Sponsored" + retailer	AI Mode users in Search
Gemini app	None	Not selling	N/A	N/A	N/A
Copilot	Perf Max conversational placement	Open via Microsoft Advertising	Standard Microsoft Advertising floors	"Microsoft Advertising" + "Sponsored"	Copilot users
Claude	None	Not selling (pledged)	N/A	N/A	N/A
Thrad (independent)	Prompt-triggered cross-surface	Open today	Transparent CPM, no six-figure floor	"Sponsored" + publisher attribution	Cross-surface prompt-intent audience

Which format fits which buyer objective?

Not every AI assistant ad is useful for every buyer. Match the format to the business outcome. A few mappings that hold up under scrutiny:

Performance ecommerce with SKUs → ChatGPT shopping cards or Copilot Perf Max

If you have a product feed, sell direct-to-consumer, and want attribution on last-click revenue, ChatGPT shopping cards (via the Ads Manager pilot if you can clear the $50K floor) or Copilot Performance Max (via your existing Microsoft Advertising account) are the two cleanest options. Both integrate with retailer checkout, both report click-through attribution, both scale on feed quality.

Brand consideration for high-consideration categories → Conversational banner (ChatGPT) or prompt-triggered cross-surface (Thrad)

For categories where the user enters the assistant already researching a multi-step decision — B2B software, financial products, healthcare services — a conversational banner in ChatGPT or a cross-surface prompt-triggered network placement is the better fit. The reason: conversational context compounds. When the user has already framed their decision in natural language, a follow-up ad unit can respond to the specific constraint, which keyword targeting cannot.

Publisher-native answer visibility → Licensing + brand-citation plays (cross-assistant)

If your goal is appearing inside answer citations across ChatGPT, Perplexity, Copilot, and Gemini answers, the format is not an ad auction — it is publisher licensing deals plus high-quality content your brand sponsors or produces. Independent networks that route brand placements through publisher surfaces, including Thrad's publisher-network integration, reach all four assistants without having a direct sales relationship with any of them.

Reach at SMB budget → Independent network (Thrad)

The direct-platform minimums — $50K at ChatGPT's pilot, Google Ads' practical floors for competitive AI Mode inventory, Microsoft Advertising's practical Performance Max floors — price most SMBs out of direct buying. An independent network is the realistic path. Transparent CPM, cross-surface reach, no waitlist.

How do measurement models differ across these surfaces?

Platform measurement practice varies widely. ChatGPT's ads manager pilot exposes impressions, clicks, and conversions via Etsy/Shopify integrations; it does not yet expose a standard third-party verification interface comparable to Google's or Microsoft's. Copilot rides on Microsoft Advertising's mature measurement stack — DV, IAS, MOAT can all verify inventory through existing integrations. Gemini AI Mode rides Google Ads' measurement infrastructure. Perplexity's measurement was early-stage and remains paused alongside its advertiser intake.

Independent networks that sell cross-surface have to meet or exceed the most mature verification set to win plans; the mature ones publish open CPM, MRC-aligned viewability, and impression-level reporting. This is the measurement posture Thrad ships by default — not because it's easy, but because the alternative (opaque vendor metrics) is a 2026 buyer red flag.

What are the common misconceptions about AI assistant ad formats?

Six misconceptions circulate in 2026 buyer conversations. Each fails under scrutiny.

"ChatGPT ads will have display banners." They do not. The two
live formats are shopping cards and conversational banners — both
native to the conversational UX. No banner ads, no interstitials,
no pre-roll.
"Perplexity is the easiest AI ad surface to buy today." It was,
until late 2025. New-advertiser intake is paused. A buyer cannot
get onboarded onto Perplexity directly as of April 2026.
"Gemini app ads are rolling out in 2026." Google publicly
denied this after its own advertiser briefings leaked. AI Mode in
Search has sponsored shopping; the Gemini app does not.
"Copilot ads require a new campaign setup." They do not. Any
existing Microsoft Advertising Performance Max campaign is
automatically eligible for Copilot placements.
"Claude will eventually ship ads." Anthropic has been
unusually direct: no. The Super Bowl campaign was a public
commitment, not a trial balloon. Revenue structure (API +
enterprise) supports the pledge.
"An independent AI-assistant ad network can't reach the same
users as direct platform buys." It can — the underlying
inventory is either the assistant's surface itself (for partner
integrations) or the web-publisher fringe that feeds citations,
which is where a meaningful share of assistant-driven user
outbound clicks resolves anyway.

What comes next for AI assistant ad formats in 2026-2027?

Four developments are likely within the next 18 months. First, ChatGPT's ads manager drops its minimum further. A $50K floor is a pilot-mode number — the economics push toward a much lower entry point within the year as OpenAI transitions from managed program to self-serve auction. Second, Perplexity relaunches or is replaced. Either the pause ends with a redesigned ad product, or Perplexity's inventory effectively migrates to independent-network fulfillment. Third, Gemini app monetization ships in some form — probably shopping before display, probably labeled, probably via the Google Ads stack. Fourth, measurement standardization. IAB Tech Lab has been working on interim guidance for generative-surface advertising, and 2026-2027 is when MRC-equivalent standards become the default RFP requirement across all AI assistant inventory.

One non-development: Claude ads will not ship in this window. Anthropic's revenue trajectory ($4.5B in 2025, per Axios) and its public positioning both point against it. A buyer planning 2026 and 2027 media should assume Claude is ad-free and plan their cross-assistant reach around the other four surfaces plus independent networks.

How do labeling and disclosure rules compare across surfaces?

Labeling is the closest thing the AI-assistant ad category has to a settled design pattern in 2026, and it varies meaningfully surface by surface. The IAB Tech Lab released interim guidance for generative-surface advertising in 2026 that pushes toward visible "Sponsored" tags with source attribution — every platform operator has implemented that standard, but with small differences that matter to brand teams running compliance review.

ChatGPT uses a short "Sponsored" label above the product name on every card and a matching disclosure inside conversational banner flows. The label is server-enforced, which means advertisers cannot suppress or modify it. Copilot's format uses "Microsoft Advertising" plus "Sponsored" as a paired tag, which is a stricter disclosure than display norms and aligns with Microsoft Advertising's existing transparency posture. Gemini AI Mode uses a standard Google "Sponsored" tag on shopping placements, consistent with the Shopping ad convention. Perplexity's paused format used a "Sponsored" tag inside its People-Also-Ask layout. Independent networks are required by contract with their inventory partners to match or exceed the strictest label on any surface the ad runs on — which in practice means "Sponsored" plus a publisher attribution string.

The implication for buyers: the label is non-negotiable on every surface. Advertisers who show up to AI-assistant inventory expecting display-era latitude on disclosure (small print, native-but-not-labeled patterns) will fail review on every surface. Treat labeling as a feature of the format, not a compliance cost.

What are the cross-surface measurement traps to avoid?

Cross-surface measurement is where most 2026 AI-assistant campaigns quietly lose value. Three traps recur:

Referrer blindness. AI assistant click-throughs often arrive
on landing pages with stripped or reduced HTTP referrer headers,
which means standard analytics under-attribute the traffic to its
source. The fix: UTM parameters on every ad destination URL plus
server-side conversion event tagging that does not depend on
referrer.
Per-platform conversion definitions. Each platform defines
conversion slightly differently — OpenAI's Ads Manager pilot
reports Etsy/Shopify-attributed conversions; Copilot reports
Perf Max conversions; Gemini AI Mode reports Shopping
conversions. Without a unified definition on your side, the
platform metrics will not aggregate cleanly.
Double-counting across assistants. A user who sees your ad
on both ChatGPT and Copilot in the same week can be counted
twice by each platform. Without server-side deduplication, the
unified view over-reports reach and under-reports true
incremental conversion.

The independent-network answer to these traps is to provide a single attribution plane that spans the surfaces it serves, with consistent conversion definitions, deduplicated reach, and server-side event logging. Thrad's measurement layer is built to address each of the three traps — because a buyer who cannot aggregate cross-surface performance will not scale spend, which hurts the network as much as the advertiser.

A fourth trap worth naming: the "last-click-is-dead" failure mode. When AI-assistant surfaces drive the upper-funnel consideration phase, last-click attribution sends the conversion credit to whichever channel closes the click — which is almost never the AI-assistant surface. Buyers who grade AI-assistant performance on last-click will under-report the channel's real contribution by a wide margin. Multi-touch attribution, holdout testing, and incrementality studies are the honest ways to grade this inventory. An independent network that publishes holdout-based incrementality reporting (rather than last-click vendor numbers) is measurably more defensible inside finance-review conversations, which is where most AI-assistant budgets actually get approved or cut in 2026.

How to get started buying AI assistant ads right now

The practical checklist for an April 2026 buyer:

Audit your product fit by assistant. If you have a product feed
and DTC ecommerce, ChatGPT shopping cards and Copilot Perf Max are
priorities. If you sell B2B services, conversational banners and
prompt-triggered cross-surface units outperform shopping formats.
If you're publisher-driven or PR-led, the citation path matters
more than the auction path.
Decide on your spend floor. If you can clear $50K commit on
ChatGPT, apply to the ads manager pilot. If not, route through an
independent AI-assistant ad network like Thrad that sells
transparent CPMs without a six-figure minimum.
Activate your Microsoft Advertising account. Copilot placements
are automatic on Performance Max campaigns. If you're not already
running Microsoft Advertising, you are leaving the most mature
AI-assistant inventory on the table.
Apply for Gemini AI Mode via Google Ads. Shopping campaign
inclusion in AI Mode is the access path — no separate Gemini buy
exists in April 2026.
Plan around Perplexity and Claude as unbuyable. Until
Perplexity reopens new-advertiser intake, do not assume Perplexity
reach. Until Anthropic reverses its stance (which it has publicly
committed not to do), Claude is off the table.
Measure across surfaces consistently. Do not let each platform
dictate its own measurement — unify on impressions, clicks,
citations, and incremental lift. An independent network makes this
easier; a platform-by-platform buy makes it harder.

The short version: ChatGPT is gated but open; Perplexity is closed; Gemini is Search-only via AI Mode; Copilot is quietly the most mature; Claude is not for sale. The through-line for a buyer who wants reach across all of it today — without waitlists, six-figure minimums, or opaque measurement — is an independent AI-assistant ad network. That's the pitch for Thrad, delivered honestly: not because the platforms are bad, but because they are not yet unified, and plan-level continuity matters more than any single platform's pitch. Ad formats will keep multiplying; the buyers who win 2026 and 2027 are the ones who pick the formats already shipping and buy them through a path they actually have access to today.