LLMrefs Review 2026: One Tier at $79, Ten Engines, and Two Different Answers About How Many Countries

LLMrefs sells one plan. Not an entry tier with a ladder above it: one plan, $79 a month, 500 prompts, every engine included, unlimited projects and unlimited seats. In a category whose median entry tier buys three engines, that is the most aggressive per-engine price we have recorded outside a free tier.

It is also a page that gives two different answers to the question of how many countries it covers, roughly four hundred words apart.

Disclosure: EchoWi sells AI visibility measurement, so LLMrefs is a competitor. Everything below about price, engines, limits and refresh frequency was read from llmrefs.com on 8 August 2026 and is quoted from that page. Demand figures come from our own keyword measurement with country and language set explicitly. We have not used the paid product, and the section on what we could not verify says so plainly.

The short version

1. One tier, $79 a month, and no ladder. “All in One / For marketing teams / $79 /month / Limited time only”. A 7-day free trial, “Cancel anytime. No questions asked!”
2. 500 prompts, all engines, unlimited projects and seats. “Track 500 prompts for brand mentions, sources and fan-out queries” and “Access all AI search engines with no additional fees”.
3. Ten engines named, including two that most of the category does not cover.
4. The page contradicts itself on geography. The marketing copy says “20+ countries and 10+ languages”. The pricing card says “50+ countries” and “20+ languages”. Same product, same page, same day.
5. “Real-time” and “at least once per week” appear in the same answer. Those describe different products to a buyer.
6. It is genuinely cheap per engine. $7.90 per engine at entry against a category median of $23.

What it costs, verified

Read from llmrefs.com/#pricing on 8 August 2026:

The engines, named in the page’s own FAQ: “We track all major AI models including ChatGPT GPT-5, Google AI Overviews, AI Mode, Gemini, Perplexity, Claude, Grok, Copilot, Meta, DeepSeek.” That is ten.

One note on how we read the price, because it matters for anyone repeating this. The URL llmrefs.com/pricing is not a page. It is a nine-byte redirect to /#pricing, an anchor on the single-page site. Everything above therefore comes from one document. We mention this because we nearly reported a contradiction between “the homepage” and “the pricing page” before checking, and there are not two pages to contradict each other.

The arithmetic nobody does

$79 a month for 500 prompts is $0.158 per prompt per refresh cycle.

The refresh commitment is “at least once per week”, so a month is about 4.33 cycles. That gives 500 × 4.33 = 2,165 prompt-runs a month, and $79 ÷ 2,165 = $0.036 per prompt-run.

Spread across the ten named engines, one refresh of the full prompt set is 5,000 prompt-and-engine observations, so $79 buys about 21,650 observations a month, or $0.0036 each.

Those are the numbers to carry into any comparison, because per-engine pricing is where this tool actually sits apart. Against the 55 tools in our catalogue that publish both an entry price and an entry engine count:

LLMrefs charges about the median price of the category for more than three times the median engine coverage. On price per engine it is tenth cheapest of fifty-five, behind two free tiers, then Rankscale at $2.00, MaxAEO at $4.75, Promptmonitor and Radarkit at $4.83, Rank Prompt at $6.50, Geoptie at $6.83 and Otterly at $7.25, and level with Qwairy at €7.90.

For contrast, the dear end of the same catalogue: Profound at $99 for one engine, Goodie at $399 for three, seoClarity’s ArcAI at $2,500 for nine.

Where the page disagrees with itself

Two figures for the same capability, roughly four hundred words apart in one document:

A buyer in Austria or Colombia cannot act on that. Both numbers are floors, so they are not strictly contradictory in logic, since a product covering fifty countries also covers twenty. But they are written as coverage claims in the two places a buyer looks, and the smaller pair appears in the more prominent one. If the real number is fifty, the marketing copy is understating the product by a factor of two and a half. If it is twenty, the pricing card is overstating it.

The second tension is about time. The FAQ answer reads in full: “We check AI search results in real-time. Prompts are continuously checked to ensure statistical significance. Every keyword is updated in the dashboard at least once per week.”

Those two sentences can both be true. A system can sample continuously and surface an aggregate weekly, and honestly, continuous sampling with weekly aggregation is a defensible design and better statistics than a single daily draw. But “real-time” is the word in the headline and “at least once per week” is the commitment, and a buyer comparing against a competitor that promises daily collection needs to know which one is the guarantee. It is the weekly one.

We flag this pattern often enough that it is worth naming: the interesting number in this category is almost never the headline. It is the floor in the small print.

Where LLMrefs is the right choice

Being fair about this matters more than being clever, so:

• You want many engines and you are price-sensitive. Ten engines at $79, including Grok and DeepSeek, is genuinely hard to match. Most of the catalogue’s $79 tier buys three or four engines.
• You are an agency or a team with several brands. Unlimited projects and unlimited seats on a single $79 subscription is unusual. Several competitors charge per seat or per domain, and that is where their bills grow.
• You want to try before committing. Seven days free with no card language on the button, and an explicit cancel-anytime statement.
• Weekly cadence is enough for you. For most brands measuring a slow-moving category, weekly is not a compromise. It is the correct sampling rate, and paying for daily collection you do not read is a common waste.

Where it is the wrong choice

• You need a daily or intra-day guarantee. The commitment is weekly. If your category moves on news cycles, this is the wrong instrument regardless of what “real-time” suggests.
• You need a specific country confirmed before you buy. Given the 20-versus-50 discrepancy, ask them in writing which countries, and get the answer before paying.
• You need more than 500 prompts and want predictable cost. The page says additional keywords can be purchased “at any time” but does not publish what they cost. That is an unpriced variable in your budget.
• You need documented methodology. Nothing on the page states how many runs per prompt back a reported number, and with one plan there is no tier where that becomes visible.

What we could not verify

• The product itself. We did not buy a subscription. Everything here is what the vendor publishes, checked against its own page.
• “Trusted by more than 10,000 marketers.” No source, no date, no definition of what counts as a marketer or whether these are paying accounts. We report the claim and attribute nothing to it.
• Whether all ten engines are covered at the $79 tier. The FAQ names ten for the platform and the pricing card says “all AI search engines with no additional fees”, which reads as ten, but the card does not enumerate them. Our per-engine arithmetic assumes ten, and if the real entry number is lower, that $7.90 rises proportionally.
• The cost of extra prompts. Absent from the page.
• Which country count is correct. Absent, in the sense that the page gives two.
• Run counts behind any figure the product reports. Absent from the page.

How it compares

Every price in that table was read from the vendor’s own page and carries a verification date in our tools catalogue. Engine counts are entry-tier counts, not platform totals, because the platform total is the number vendors like to print and the entry count is the one you get for the money.

Is the demand real?

We check this before writing any review, because a review of a vendor nobody searches for is not worth publishing.

“llmrefs” returns 260 searches a month in the United States with a keyword difficulty of 10 and a cost per click of $34.61, measured on 8 August 2026 with country and language set explicitly. The term had no measurable volume at all before March 2025.

The cost per click is the interesting part. Thirty-four dollars for a single click on a brand name is not a curiosity term. Someone is bidding real money against buyers who type this name, which is the strongest available signal that the searches convert. Compare our own note on this method: a low cost per click beside an informational intent is how we detect that a “brand” term is really a common word. This is the opposite pattern.

Common Questions About LLMrefs

How much does LLMrefs cost?

One plan at $79 a month, labelled “All in One” and marked “Limited time only”, read from llmrefs.com on 8 August 2026. There is no cheaper tier and no published tier above it. A 7-day free trial is offered with “Cancel anytime” wording.

How many AI engines does LLMrefs track?

Its FAQ names ten: ChatGPT GPT-5, Google AI Overviews, AI Mode, Gemini, Perplexity, Claude, Grok, Copilot, Meta and DeepSeek. The pricing card does not list them individually, saying instead “Access all AI search engines with no additional fees”.

How often does LLMrefs refresh its data?

The page says “We check AI search results in real-time” and, in the same answer, “Every keyword is updated in the dashboard at least once per week”. The commitment a buyer can hold them to is the weekly one.

How many countries does LLMrefs cover?

The page gives two answers. Its marketing copy says “20+ countries and 10+ languages”; its pricing card says “Geo-targeting in 50+ countries” and “20+ languages”. We would ask in writing before buying if a specific market matters to you.

Is LLMrefs good value?

On price per engine, yes, measurably. At $79 for ten engines it costs $7.90 per engine against a category median of $23 across the 54 catalogued tools that publish both figures, which puts it tenth cheapest of fifty-four. Whether that is good value for you depends on whether a weekly refresh and 500 prompts fit your category, and on getting the country answer confirmed.