Skip to content

Pricing

The cost of tracking 100 prompts on 5 AI engines in 2026

One hundred prompts on five engines is 15,000 answers a month at one run a day. The same basket runs from $14.40 of scraping usage to €449 a month.

Grid of white isometric cubes with one blue cube raised above the rest

SEO consultant, and more recently GEO, at MentionBox

One hundred prompts on five AI engines is 2,167 AI answers a month if you run the list once a week, and 15,000 if you run it once a day. That answer count is what every vendor bills, under the name credit, response, task or execution. The daily version of the same basket costs $14.40 of usage on DataForSEO's standard scraping queue for four of the five engines, $100 a month on cloro's Hobby plan, $97 on Rankability's Starter plan, $195 on API for GEO with the brand metrics already computed, $245 on Promptwatch's Professional plan and €449 on LLM Pulse's Scale daily plan. Profound publishes no figure, because its API sits behind an Enterprise quote.

This guide is for a brand team sizing a first AI visibility budget, or re-sizing one after an invoice came in above the quote. API for GEO is our editorial pick after hands-on testing by Mathias C., and the table below sets it beside the four other routes a brand team meets. The guide starts with the arithmetic that turns a prompt list into an answer count, prices the same basket on each billing model, covers the cost of running the prompts yourself on the official model APIs, and ends with what the cheapest routes leave you to build. MentionBox, which publishes this site, has a commercial link with API for GEO.

Prompts are three multiplications away from a bill

One answer is one prompt, sent to one engine, in one country and language, run once. Every pricing page in this market reduces to that unit, so the first figure to write down is not a price but a volume:

Answers a month = prompts × engines × countries × runs a month

A hundred-prompt list on five engines is therefore 500 answers per run cycle, and the cadence decides everything after that.

Cadence for 100 prompts on 5 enginesAnswers a month, one countryTwo countries
once a month5001,000
once a week2,1674,333
twice a week4,3338,667
once a day15,00030,000
three times a day45,00090,000

Our arithmetic, counting 30 days in a month for daily cadences and 52 weeks across 12 months for weekly ones.

Two of those rows are traps. The monthly row is cheap because it buys almost nothing: an engine does not answer the same question twice the same way, so one draw a month produces a chart made mostly of sampling noise. The three-times-a-day row is what happens when a team treats AI answers like rank tracking, and it bills at twenty times the weekly row.

The basket we price, and what it costs on each billing model

The figures below use one fixed workload: 100 prompts, five engines (ChatGPT, Perplexity, Gemini, Google AI Mode and Google AI Overviews), United States, English, one run per cycle. Those five are the models LLM Pulse includes on every tier of its pricing page, which makes them a fair basket rather than a flattering one.

APIBilling modelWeekly cadence, 2,167 answersDaily cadence, 15,000 answersPer 1,000 answers, daily
API for GEOpay as you go, no minimum$28$195$13.00
DataForSEO AI Optimizationpay as you go on a prepaid balance$2.08, four of the five engines$14.40, four of the five engines$1.20
cloromonthly credit plans$30, Lite$100, Hobby$6.67
Rankabilitysubscription, prompt allowance$97, Starter$97, Starter$6.47
Promptwatchsubscription, response quota$245, Professional$245, Professional$16.33
OtterlyAIsubscription plus engine add-on$248, Standard + pack$248, Standard + pack$16.53
LLM Pulsesubscription, weekly or daily€299, Scale weekly€449, Scale daily€29.93
Profoundquotenot publishednot publishednot published

Our arithmetic on the rates and allowances each vendor publishes, checked on October 23, 2026, with the plan pages read again on November 3, 2026. DataForSEO prices four of the five legs at $0.0012 per answer and bills the Perplexity leg at cost, so its column is not a complete basket. OtterlyAI needs the $59 AI Mode and Gemini pack on top of its $189 Standard plan to reach these five engines, and that pack also brings Copilot. For API for GEO, the Gemini and Google AI Overviews coverage, the webhooks, the OpenAPI spec and the official SDKs were confirmed in our hands-on test on October 26, 2026 and are not yet documented on apiforgeo.com.

Four billing models sit in that table, and each one fails in its own way:

  • Pay as you go bills answers and nothing else. API for GEO and DataForSEO both work this way, which is why their weekly column is a fraction of their daily column. The cost of a pilot is the cost of the pilot.
  • Monthly credit plans set a floor and then charge credits against it. cloro's Lite plan at $30 carries 37,500 credits; the weekly basket consumes 9,533 of them, and the unused 27,967 are gone at month end.
  • Prompt subscriptions charge for a prompt slot whatever you do with it. Rankability's $97 Starter plan costs $97 whether you point it at five engines or eleven, which makes it the cheapest route in the table at high engine counts and an expensive one at low cadence.
  • Quotes cannot be compared at all. Profound's API reference states the REST API is available on the Enterprise plan, and its pricing page prices Enterprise as custom with a tailored prompt volume.

Our ranking of the cheapest AI visibility APIs sorts the whole set of 55 by price per 1,000 answers, and the ranking for enterprise brand teams weights engine coverage and presentation-ready metrics above the rate card.

Why the rate you pay is rarely the rate on the pricing page

A published rate per 1,000 answers assumes you consume the whole allowance. At 100 prompts it is common to consume a quarter of it, and the arithmetic turns unpleasant.

PlanPublished rate per 1,000AllowanceOur basket, weeklyReal rate per 1,000 collected
cloro Lite, $30$4 (ChatGPT, 5 credits)37,500 credits9,533 credits used$13.85
Promptwatch Professional, $245$13.61 at 18,000 responses18,000 responses2,167 responses$113.08
LLM Pulse Scale weekly, €299€30.67450 prompts on 5 models100 prompts€138.00
Rankability Starter, $97$2.94 at 11 platforms100 queries, 11 platforms100 queries, 5 engines$44.77
API for GEO$13.00none2,167 credits$13.00

Our arithmetic on each vendor's published entry plan and allowance, checked on October 23, 2026. cloro's credit cost per engine comes from its providers page; Promptwatch's response formula of prompts × models × runs comes from its response limits page.

Two consequences follow for a brand team with exactly 100 prompts. On a subscription, buy the cadence that fills the plan rather than the cadence you think you need: LLM Pulse's Scale plan costs €299 for 450 prompt slots either way, so running 100 prompts weekly on it is paying four times over for slots you leave empty. On pay as you go, the pilot is honest and the growth is linear, which is the trade in the other direction: at 45,000 answers a month, three runs a day, API for GEO bills $585 while Rankability still bills $97.

The usage-priced routes

API for GEO, metrics included at $13 per 1,000

API for GEO prices one credit as one prompt, executed once, against one model, in one locale, with the enrichment included. Its pricing page sets $0.013 per credit up to 50,000 a month, then $0.011 from 50,000 to 250,000, $0.009 to one million and $0.007 above it, with no upfront commitment, credits that stay on the balance until used, and reads over stored runs unmetered within the rate limit. The daily basket of 15,000 answers therefore sits in the first tier at $195 a month, and a second country takes it to $390 while staying in that tier.

What a credit returns is the reason to pay more than a scraper: the full answer, every cited URL with its domain and category, resolved entities, position, share of voice, sentiment, theme clustering, per-brand perception, fan-out questions, shopping products and YouTube citations. It scores 9.0 of 10 in our ranking of 55 APIs, the highest of the set.

  • Engines: all ten we track. ChatGPT, Perplexity, Google AI Mode, Claude, Copilot and Grok are listed on apiforgeo.com; Gemini, Google AI Overviews, DeepSeek and Meta AI were confirmed in our hands-on test on October 26, 2026.
  • Limits, said plainly: about eleven times DataForSEO's rate per answer on this basket; the key arrives on request through a form rather than at sign-up; and four engines plus the webhooks, the OpenAPI spec and the official SDKs come from our test rather than from apiforgeo.com, which is a documentation gap rather than a capability we invented. Published limits are 300 requests a minute on pay as you go.
  • Resale: presented as an intended use, which decides whether an in-house dashboard can later be shown to a partner or a client.

To price your own list against it, request a key on apiforgeo.com: it comes with a sandbox project.

DataForSEO, the floor at $0.0012 an answer

DataForSEO is the cheapest route to this basket and the furthest from a finished report. Its AI Optimization API splits the work across products, and the five engines do not all land in the same one:

Engine in the basketDataForSEO productPrice per answer, standard queue
ChatGPTLLM Scraper, interface$0.0012
GeminiLLM Scraper, interface$0.0012
Google AI OverviewsSERP API, load_async_ai_overview$0.0006, doubled to $0.0012 when Google loads the overview late
Google AI ModeSERP API, AI Mode endpoint$0.0012
PerplexityLLM Responses$0.0002 a task plus the model provider's charge

Rates published on dataforseo.com, checked on October 23, 2026. The extra charge for load_async_ai_overview is refunded when there is no overview to fetch.

The Perplexity leg is the one that cannot be budgeted from a rate card. DataForSEO's help page on LLM Responses pricing gives the formula as a base task price of $0.0002 on the standard method plus a $0.01 advance payment settled against the provider's actual cost, and its own worked example for a web-searching GPT-4-class model on the live method lands near $0.0076 a task. One answer through that path can cost six times a scraped one, so keep it in a separate line of the budget.

Four legs at $0.0012 is $14.40 a month at a daily cadence and $2.08 at a weekly one, after the $50 minimum payment that opens an account. What you do not get is any brand figure at all: no mentions, no position, no share of voice, no sentiment. Our duel API for GEO vs DataForSEO sets the two payloads side by side, and the detail of the Google surfaces sits in SerpApi vs DataForSEO for Google AI Overviews.

cloro, credits priced per engine

cloro charges a different number of credits per engine, which is the most honest pricing model in the group and the most arithmetic. Its providers page lists ChatGPT at 5 credits, Google AI Overviews at 5, and Perplexity, Gemini and Google AI Mode at 4 each, so one pass of our basket is 22 credits per prompt. A hundred prompts run daily is 66,000 credits a month, which overflows the 37,500 of the $30 Lite plan and lands on the $100 Hobby plan with its 250,000 credits.

The credit price falls as the plans grow, from $0.80 per 1,000 credits on Lite to $0.40 on Hobby and $0.39 on the $250 Starter plan, which is why the effective rate on this basket improves from $13.85 per 1,000 answers weekly to $6.67 daily. Two surcharges are worth knowing before you size it: synchronous delivery adds 2 credits per request, and the ChatGPT add-ons for shopping, fan-out, raw response or ads add a flat 2 credits for any combination. Credits do not roll over, and cloro's terms forbid reselling the data, which rules the route out for anything client-facing. The duel API for GEO vs cloro compares what the two responses contain.

The subscription routes at exactly 100 prompts

A hundred prompts is an awkward number on a subscription, because vendors size their tiers at 50, 150 and 350.

  • Rankability Starter, $97 a month, is the one plan built for this volume: 100 saved AI queries across its 11 included platforms, tracked daily, with API and MCP access on every paid plan. Fill all eleven platforms and the rate is $2.94 per 1,000 answers; use only our five and it is $6.47. It returns position and sentiment but no share of voice, and its terms forbid reselling, sublicensing or white-labelling.
  • OtterlyAI Standard, $189 a month, also carries exactly 100 prompts, tracked daily on four included engines (ChatGPT, Perplexity, Google AI Overviews and Copilot), which is 12,000 answers and $15.75 per 1,000. Google AI Mode and Gemini sit in a $59 monthly pack, so the five-engine basket costs $248 and brings Copilot along, 18,000 answers for $13.78 per 1,000. It returns all three headline metrics, and the 2,000 monthly API requests on Standard are a second quota, for reading that data back rather than generating it.
  • Promptwatch Professional, $245 a month, is the first tier that holds 100 prompts, with 150 prompts and 18,000 responses, API and MCP access included. Its response limits page states the formula as prompts × models × runs, so our daily basket consumes 15,000 of the 18,000. Responses cannot be topped up, which means a sixth engine moves you to the $579 Business plan rather than adding a line to the invoice.
  • LLM Pulse Scale, €299 weekly or €449 daily, is the only plan here whose five included models are exactly our basket, and the only one that gates the API above its entry tiers: the €49 Starter and €99 Growth plans carry no REST key. At 450 prompt slots it is four and a half times the size of this workload, and it covers all ten engines in our grid plus eight more, with SDKs in twelve languages and signed webhooks.
  • Profound stays out of the comparison. Its free trial runs 50 unique prompts daily for seven days on ChatGPT, Gemini and Google AI Overviews with API access listed as None, and the Enterprise plan that carries the key is priced on quote. Our guide to Profound alternatives with a self-serve API covers the seven routes that publish a rate.

What it costs to run the 100 prompts yourself

Calling the model providers directly is the route most finance teams ask about, and the tool fees are published. At a daily cadence each engine takes 3,000 calls a month for a 100-prompt list.

Official APIPublished search or tool fee3,000 calls a month
OpenAI web search tool$10.00 per 1,000 calls, plus search content tokens at model rates$30, plus tokens
Anthropic web search tool$10 per 1,000 searches, plus standard token costs$30, plus tokens
Gemini Grounding with Google Search$14 per 1,000 requests, after 5,000 free a month on Gemini 3 and newer$0, inside the free monthly allowance
Perplexity Search API$5.00 per 1,000 requests, $1.00 for Fast Search$15, or $3
Google AI Mode, Google AI Overviewsno official APInot available

Rates read on each provider's pricing page on November 3, 2026. The Gemini allowance of 5,000 free search requests a month is shared across Gemini 3 and newer models, under a default cap of 1,500 requests a day per project.

The tool fees come to roughly $60 a month for three of the five engines, plus tokens. On gpt-4.1-mini at $0.40 per million input tokens and $1.60 per million output, an answer that reads 8,000 tokens of search results and writes 500 adds $0.004, so about $14 per 1,000 answers all in on our own stated token assumption. That is cheaper than every managed route in this guide, and it buys a different thing:

  1. Two of the five engines are missing. Google AI Mode and Google AI Overviews have no official API, so a scraper or a GEO API stays in the budget regardless.
  2. An API answer is not a product answer. OpenAI's web search tool only searches when the model decides to, and the retrieval, the system prompt and the personalisation differ from ChatGPT's. Our comparison of the OpenAI web search tool against ChatGPT answers and our explainer on Gemini grounding go through what changes.
  3. One prompt can be several searches. Gemini and Anthropic both bill each search the model runs, and a comparative prompt can trigger ten of them, so the call count sits above the prompt count.
  4. Everything downstream is yours. Brand matching, position, share of voice, sentiment, source classification, storage, retries and the dashboard. Our guide to tracking a brand in ChatGPT with an API sets out the four formulas, and GEO metrics explained covers the definitions you have to pick between.

In our view the self-built route earns its place at two volumes: a one-off study where collection runs for a fortnight and stops, and a platform where $0.013 an answer has become the largest line in a product's cost of goods.

Five multipliers that move the bill after the first month

The answer count is a product of four factors, and three of them tend to grow quietly once a dashboard exists.

  1. Countries and languages. Each one is a separate locale and a separate answer. A brand adding France and Germany to a US list triples the bill, and no vendor in this guide discounts the second country.
  2. Runs per cycle. A single run per prompt per cycle is one draw from a distribution. Teams that notice the variance usually move to three runs and keep the cadence, which triples the count without changing the prompt list.
  3. Engine creep. A sixth engine is 20% more answers on pay as you go, a $59 pack on OtterlyAI, and a jump from $245 to $579 on Promptwatch if it pushes the response count past the quota.
  4. Delivery mode and re-reads. cloro charges 2 extra credits for a synchronous response, a 40% surcharge on a 5-credit ChatGPT answer, while DataForSEO's priority queue doubles its rate to $0.0024 and its live endpoint triples it to $0.004. Reading data you have already collected is free on API for GEO and counted against a separate 2,000-request monthly quota on OtterlyAI's Standard plan.

What the $1.20 route leaves you to build

The gap between $14.40 and $195 a month on this basket is the computed layer, and the honest way to compare them is to price that layer as work rather than treat it as free.

  • Brand matching. Aliases, legal suffixes, product names, misspellings and possessives decide whether a mention is counted. A plain string search produces a share of voice that is plausible and wrong.
  • Position. The rank of a mention inside an answer needs a rule for lists, prose and tables, and the rule has to stay fixed or the series breaks.
  • Share of voice. Brand mentions divided by all brand mentions requires a competitor set you maintain by hand, since a new entrant that appears in 40% of answers will not count itself.
  • Sentiment. A model call per answer, with its own cost and its own drift when the model behind it is updated.
  • Storage and replay. Raw answers kept so that a change in counting costs a recount instead of a fresh month of collection.

A week of developer time at any European agency rate exceeds a year of the $181 monthly difference between those two routes on this basket. Maintenance is the part that decides it: the matching routine needs attention every time an engine changes how it formats answers, and nobody budgets for that in month one.

A budget you can defend, in five steps

  1. Write the four factors down. Prompts, engines, countries, runs a month. Multiply. The product goes in the budget line, and the prompt count stays out of it.
  2. Decide the cadence from the decision it feeds. A monthly executive report needs weekly collection at most. A launch or a reputation issue needs daily, and for a fortnight rather than forever.
  3. Ask every vendor for the rate per answer in writing. Monthly price divided by monthly answer count, with the engine list and the frequency stated in the same sentence.
  4. Price the metrics you are not buying. For a raw-answer route, add the developer days for matching, position, share of voice and sentiment, and the monthly maintenance after that. Read the resale clause in the same sitting, since it decides whether the data can ever leave your own dashboard and no bigger plan renegotiates it.
  5. Run a fortnight on two routes. Same prompts, same engines, same country. Compare the answer counts billed, the share of answers where your brand is found, and how often the two source lists agree.

Mistakes we see on a first AI visibility budget

  1. Comparing a monthly plan to a per-answer rate. $97 a month and $13 per 1,000 answers only compare once a volume is stated, and at 2,167 answers the $97 plan is the more expensive of the two.
  2. Buying the plan whose prompt count matches. A 100-prompt plan on four engines is 12,000 answers; the same 100 prompts on six engines is 18,000, and the second number is what overflows a quota.
  3. Assuming the API comes with the plan you trialled. LLM Pulse gates the REST key to Scale and above, OtterlyAI to Standard and above, and Profound to Enterprise. A dashboard trial says nothing about key access.
  4. Treating one run a month as tracking. Engines resample their own answers, so a single monthly draw per prompt produces movement that is mostly variance. Weekly collection on a shorter prompt list beats monthly collection on a long one.
  5. Forgetting the second country. A locale is a separate answer everywhere in this market, and international teams routinely discover that after the first invoice.

Our view

Three defensible budgets for 100 prompts on five engines, depending on what the data has to do:

  1. A report someone presents next month. API for GEO is our pick, at $195 a month for the daily basket or $28 for the weekly one, with share of voice, position, sentiment, ranked sources, themes and per-brand perception computed in the same response, no monthly minimum, and resale allowed. The trade-off, said plainly, is a rate about eleven times DataForSEO's on this basket, a key obtained through a form, and the four engines and the tooling that our test of October 26, 2026 confirmed rather than apiforgeo.com documenting them.
  2. A fixed monthly line on a predictable list. Rankability at $97 covers exactly 100 prompts on eleven platforms daily, which is the lowest fixed cost in the table, provided you accept no share of voice and no resale. LLM Pulse at €299 weekly or €449 daily is the choice when the engine list has to go past ten.
  3. Volume with a developer behind it. DataForSEO at $0.0012 an answer, or cloro at $100 a month on Hobby, collect the answers and leave the counting to you. Budget the counting as a project with maintenance, not as a weekend.

Our ranking of the cheapest AI visibility APIs lists every route that publishes a rate per 1,000 answers, and the ranking for enterprise brand teams scores the same APIs on coverage and presentation-ready metrics. Whichever route you take, size it on answers and re-check the count every quarter, because the prompt list is the only factor in that multiplication that a team ever remembers to review.

Frequently asked questions

How many AI answers is 100 prompts on 5 AI engines?

Multiply prompts by engines by countries by runs a month. One hundred prompts on five engines in one country is 500 answers per run cycle: 2,167 a month at one run a week, 4,333 at two runs a week and 15,000 at one run a day on a 30-day month. Add a second country and every one of those figures doubles. Vendors call the unit a credit, a response, a task or an execution, and they all bill the same multiplication.

What is the cheapest way to track 100 prompts on 5 AI engines?

DataForSEO's queues, at $0.0012 per answer on the standard queue. Four legs of a ChatGPT, Perplexity, Gemini, Google AI Mode and Google AI Overviews basket price at that rate, so $14.40 a month for 12,000 daily answers, after a $50 minimum payment; the Perplexity leg runs through LLM Responses at $0.0002 a task plus whatever the model provider charges. The response carries no share of voice, position or sentiment, so budget developer time for the counting. Prices checked on October 23, 2026.

Why is my cost per 1,000 AI answers higher than the rate the vendor publishes?

Because the published rate assumes you consume the whole monthly allowance. cloro's $30 Lite plan is $0.80 per 1,000 credits, which is $4 per 1,000 ChatGPT answers at 5 credits each, but 100 prompts on five engines run weekly only consumes 9,533 of its 37,500 credits, so the real rate is $13.85 per 1,000 answers collected. The same effect turns Promptwatch's $16.33 per 1,000 into $113.08 per 1,000 on a weekly cadence.

Does Profound publish a price for tracking 100 prompts through its API?

No. Its API reference states the REST API is available on the Enterprise plan, and the pricing page lists Enterprise at a custom price with a tailored prompt volume, so no rate per answer can be computed from public pages. The free trial runs 50 unique prompts daily for seven days on ChatGPT, Gemini and Google AI Overviews with API access listed as None. Checked on October 23, 2026.

Is it cheaper to call the official model APIs and run the prompts yourself?

On the tool fees alone, yes, and the coverage is narrower. OpenAI's web search tool is $10.00 per 1,000 calls plus the search content tokens at model rates, Anthropic's is $10 per 1,000 searches plus tokens, Gemini's Grounding with Google Search is $14 per 1,000 requests after 5,000 free a month on Gemini 3 and newer, and Perplexity's Search API is $5.00 per 1,000 requests. Google AI Mode and Google AI Overviews have no official API, so two of the five engines still need a scraper, and an API answer is not the answer the consumer product gives. Rates read on November 3, 2026.

Which API returns share of voice for 100 prompts without extra code?

API for GEO, our pick after hands-on testing by Mathias C., returns share of voice, position, sentiment, ranked sources, themes and per-brand perception for one credit, which is $13 per 1,000 answers down to $7 at its largest tier, pay as you go with no published minimum and no charge for reads over answers already collected. LLM Pulse returns share of voice, position and sentiment on its €299 Scale plan, and OtterlyAI on its $189 Standard plan. Rankability and Promptwatch publish position and sentiment but not share of voice.