Now live across the AI ecosystem: ChatGPT GPT Store · MCP Registry · mcp.so

Comparisons

The entry price for AI visibility collapsed. Count the engines it buys.

By Arnav Mukherjee, founder of TofuBofu · August 28, 2026

We keep a comparison table on our pricing page, and one of its cells said Profound was demo-only. That was true for most of the time this company has existed. I went to re-check it against their own page this morning, the way we are supposed to before quoting anybody, and it is not true any more. Profound lists $99 a month.

So I checked the other two. Peec lists $80, in dollars, not the euro figure we had. Otterly lists $29. The category that spent two years hiding behind "book a demo" now publishes its floor, and the floor is roughly one dinner.

Then I read what the floor includes, and that is the actual story. Profound's $99 tier is described on their own page as "ChatGPT tracking only". One engine. The price came down and the measurement came down with it, and the second half is not on the pricing card in a font anybody reads.

What $99 buys at the category leader

Read at Profound's own pricing page on 28 August 2026, with a billed-yearly toggle offering two months free:

Give them credit where it is due: publishing the number at all is a real improvement on a demo gate, and the Growth tier is a serious instrument. This is not a complaint about Profound's quality, which is settled. It is a point about what the entry tier of this category has quietly become.

Because the jump from one engine to three engines is the jump from $99 to $399. Engine coverage is the meter now. Price is just the thing on the front of the card.

The same shape at Peec and Otterly

Peec, read on 28 August 2026, lists Starter at $80, Pro at $205 and Advanced at $420. All three of those tiers say "Choose 3 models", from a menu of ChatGPT, Google AI Mode, Google AI Overviews, Microsoft Copilot, Perplexity and Gemini. A fourth model is an add-on: $30 a month at Starter, $70 at Pro, $140 at Advanced. Claude appears only in the Enterprise column.

Otterly, same day, lists Lite at $29, Standard at $189, Premium at $489 and Enterprise from $1,000. The base tiers cover four engines. Claude is an add-on from $29 a month at Lite up to $439 at Premium, and Google AI Mode and Gemini are add-ons from $9 up to $149 each.

Three vendors, three pricing philosophies, one identical decision: the cheap tier is cheap because it watches fewer engines, and Claude is the one most often missing or most expensive to add. That is not a coincidence, and it is worth asking your vendor why.

Engines included at the entry tier Read at each vendor's own pricing page, 28 August 2026. Add-on engines not counted. Profound Starter Peec Starter Otterly Lite TofuBofu Track $99/mo $80/mo $29/mo Free 1 engine 3 engines 4 engines 6 engines

You can see where you stand on all six engines before you pay anybody, including us. The scan is free and takes about a minute.

Run a free AI visibility scan

One engine is not a cheaper version of six

The obvious defence of a single-engine tier is that ChatGPT is the biggest, so watching it is a reasonable sample of the whole. Our own data says that defence does not hold, and the reason is sources.

We recovered every source link out of our stored AI answers last week. On the 34 buying questions where both Google AI Mode and Bing Copilot produced sources, the two engines shared no cited domain at all on 12 of them, and the mean overlap across all 34 was 0.065. Google AI Mode cited 20 distinct sources per answer. Copilot cited 2.5, with a median of 3 among the answers that cited anything.

That comparison rests on four reports across three brands, which is a small sample and I will not inflate it. What makes it worth anything is that it is a true pairing rather than two unrelated studies: the same 34 question cells, in the same scans, in the same minute. Two engines asked identical questions at the same instant, and they built their answers out of almost entirely different websites.

If that is what happens between two engines, a subscription that watches one of them is not measuring 1/6th of your visibility. It is measuring a different thing, and telling you it is the number.

What to do with this

Divide the price by the engines, not by the prompts. Prompt counts are the number vendors compete on because they are cheap to raise. Engine coverage is the number that decides whether the report describes your market or one slice of it.

Ask whether the ChatGPT probe searches the web. This is the single most consequential setting in the category and it appears on nobody's pricing page. An engine answering from training weights is telling you what a model remembers, not what your buyer sees. We ran ours the wrong way for months, found it, and said so.

Check where Claude sits on the price list. On the three pages I read today it is behind Enterprise, or it is the most expensive add-on offered. Whether that matters depends on your buyers, and you cannot find out from a tool that does not measure it.

Then ask what happens after the dashboard. Every tool on this list, ours included, can tell you that you are missing. The ones worth paying for are the ones that write the page, publish it, and re-scan to show the number moved. Measurement is commoditised now, and the collapsing entry price is the proof. Doing the work is not.

Frequently asked questions

How much do AI visibility tools cost in 2026?

Entry pricing is public and low, and it moved recently. Read at each vendor's own pricing page on 28 August 2026: Otterly lists Lite at $29 a month, Standard at $189 and Premium at $489, with Enterprise starting from $1,000. Peec lists Starter at $80 a month, Pro at $205 and Advanced at $420. Profound lists Starter at $99 a month and Growth at $399, with Enterprise custom. Prices in this category change on a timescale of weeks, so treat any figure including these as dated rather than current, and check the vendor's own page before you buy.

What is the cheapest AI visibility tool?

By list price, Otterly's Lite tier at $29 a month is the cheapest paid entry point among the tools we checked on 28 August 2026. The number to compare it on is not the price, though, it is what the price measures. Otterly's base tiers cover four engines, and Claude, Google AI Mode and Gemini are paid add-ons on top, with Claude running from $29 a month at Lite up to $439 at Premium. A cheaper subscription that measures fewer engines is not a discount on the same product, it is a smaller measurement.

How much does Profound cost?

Profound's pricing page listed Starter at $99 a month and Growth at $399 a month on 28 August 2026, with Enterprise on custom pricing and a billed-yearly toggle offering two months free. This is a change worth noting if you read an older comparison: Profound's pricing was quote-only for much of its life, and a great deal of published writing about the category, including some of our own, still describes it that way. The figures are on their own page now.

Which engines does Profound track at each tier?

Profound's own pricing page states it plainly. Starter is described as "ChatGPT tracking only" with 50 prompts tracked. Growth is described as "3 Answer Engines tracked" with 100 prompts. Enterprise is described as "Up to 9 Answer Engines tracked". So the $99 entry tier at the category leader measures a single engine, and the jump to three engines is the jump to $399. Read at their pricing page on 28 August 2026.

Does Peec AI track Claude?

Only on Enterprise, according to Peec's own pricing table read on 28 August 2026. The Starter, Pro and Advanced columns offer a menu of ChatGPT, Google AI Mode, Google AI Overviews, Microsoft Copilot, Perplexity and Gemini, and every one of those three tiers is limited to three models chosen from that menu. Claude Sonnet 4 appears only in the Enterprise column. Additional models can be added to a lower tier for $30 a month at Starter, $70 at Pro and $140 at Advanced.

Is tracking one AI engine enough?

Our own data says no, and the reason is that engines do not draw on the same sources. On the 34 buying questions where both Google AI Mode and Bing Copilot produced sources in our scans, the two engines shared no cited domain at all on 12 of them, and the mean overlap across all 34 was 0.065. That comparison rests on four reports across three brands, which is small, and it is a real pairing rather than two separate samples because it is the same questions in the same scans at the same minute. If two engines answering identically worded questions barely intersect on sources, a subscription that measures one of them is not a cheaper version of measuring six.

How should I compare AI visibility tools on price?

Divide by what you are actually buying, then check three things the pricing page usually buries. First, how many engines the tier includes, and whether the ones your buyers use are add-ons. Second, whether the ChatGPT probe searches the web or answers from training weights, because that single setting changes the answer completely and appears on nobody's pricing page. Third, whether the tool only measures or also produces and publishes the fix, since a dashboard nobody has time to act on is a cost with no outcome. Price is the easiest number on the page and the least informative.

Sources and further reading

A note on what these figures are. Competitor prices here are quoted from each vendor's own pricing card as a subject of comparison, not used as evidence for how AI engines behave. Pricing in this category changes on a timescale of weeks, so every figure above carries the date it was read.

Related reading