Guides
API for GEO vs Peec AI for agencies: which API to buy
Peec AI opened its API to Pro, Advanced and agency plans on October 27, 2026. API for GEO bills $13 per 1,000 answers with no plan to buy first.

By Mathias C.SEO consultant, and more recently GEO, at MentionBox
Peec AI's API stopped being an Enterprise-only feature on October 27, 2026: its API changelog records under v0.39.0 that Customer API availability moved to Pro, Advanced, Enterprise and agency plans, and the API introduction now reads the same way. For an agency that changes the comparison, because the cheapest Peec AI plan carrying a key is Pro at $245 a month instead of a quote. Against it, API for GEO, our pick of the group after hands-on testing by Mathias C., sells the same unit at $13 per 1,000 answers with no plan to buy first.
This face-off is for an agency that reports AI visibility for several clients and wants the numbers in its own pipeline rather than in someone else's dashboard. The two products answer different questions: Peec AI is a tracking platform whose API reads the workspace it already fills, and API for GEO is an execution API you call to collect answers. Below: what each key costs per 1,000 answers, what the two payloads contain, what the resale clauses allow, and the four traps we see agencies hit on either route. MentionBox, which publishes this site, has a commercial link with API for GEO.
The two APIs side by side
| API for GEO | Peec AI | |
|---|---|---|
| What the API does | runs prompts on demand and returns the enriched answer | reads and manages a workspace the platform tracks for you |
| Cheapest plan with a key | none, pay as you go | Pro, $245 a month; agency Essential, $245 a month |
| Price per 1,000 answers | $13, down to $7 above 1 million credits a month | $24.50, $19.80 or $12.23 on the three agency plans, if filled |
| Monthly minimum | none published | the plan fee, paid whether you call the API or not |
| Engines, of the 10 we track | 10 | 9, Meta AI missing |
| Other surfaces | none published beyond the ten | OpenAI Search API, Qwen, Mistral, Muse Spark, Naver AI, Amazon Rufus |
| Brand metrics returned | share of voice, position, sentiment, themes, per-brand perception | share of voice, position, sentiment, visibility, win rate |
| Collection method | "executed natively on each engine, in that country and language", interface or API not named | "UI scraping rather than APIs for most tracked engines"; add-on models through official APIs |
| Countries | 187 on its published list | 80+ with dedicated infrastructure |
| Rate limit | 300 requests a minute, 10 concurrent executions | 200 requests a minute per project |
| Tooling | MCP server; webhooks, OpenAPI spec and official SDKs confirmed in our test, not documented by the vendor | OpenAPI spec, MCP server, batch writes up to 1,000 items; no SDK, no webhooks documented |
| Resale | allowed, named as an intended use | not allowed as resale; client delivery allowed under Peec's own agreement |
| Our score | 9.0 of 10 | 6.5 of 10 |
Published by each vendor on its pricing page and in its API reference, checked on October 23, 2026, with the Peec AI pricing page, API introduction and changelog read again on October 30, 2026. For API for GEO, four of the ten engines, Gemini, Google AI Overviews, DeepSeek and Meta AI, plus 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. Country figure for API for GEO checked on October 28, 2026.
Who can get a Peec AI key since October 27
The gate moved twice in two months, so a comparison written in September is wrong today. Three pages now agree, which is worth checking before you quote a client:
- The API introduction: "Access to this API is available on Pro, Advanced, Enterprise, and agency plans. Starter plans don't include API access."
- The changelog, v0.39.0 of October 27, 2026, under Changed: "Customer API availability now limited to Pro/Advanced/Enterprise/Agency plans."
- The pricing page, which lists the API on Pro, Advanced and Enterprise, and MCP on all four tiers including Starter.
Keys come in two scopes, documented on the authentication page: company-scoped keys reach every project in the organisation, project-scoped keys reach one. For an agency running one project per client, project-scoped keys are the ones to hand to a client-side integration, and the company key stays in your own backend. The header is x-api-key against https://api.peec.ai/customer/v1/.
One API executes, the other reads
The architectural difference decides more than the price does. Nothing in Peec AI's reference, which runs to roughly 150 endpoints across reports, project, products, actions, brand perception, agent analytics and company, executes a prompt. You create prompts, brands, tags and topics, and the platform collects on the frequency your plan pays for. What you read back is an aggregate, and a good one:
curl -X POST "https://api.peec.ai/customer/v1/reports/brands" \
-H "x-api-key: $PEEC_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"project_id": "'"$PROJECT_ID"'",
"start_date": "2026-10-01T00:00:00Z",
"end_date": "2026-10-30T00:00:00Z",
"dimensions": ["model_channel_id"],
"include_previous_period": true,
"having": [
{ "field": "brand_id", "operator": "in", "values": ["kw_abc123"] }
]
}'
The brands report publishes the formula behind each metric, which is rarer than it should be: share of voice is mention_count / sum(mention_count), visibility is sum(visibility_count) / sum(visibility_total), position is sum(position_sum) / sum(position_count), and sentiment is ((sum(sentiment_sum) / sum(sentiment_count)) / 2 + 0.5) * 100. Win rate, sum(win_count) / sum(visibility_total), has no equivalent in the other payload. Dimensions available are prompt, model, model channel, tag, topic, date, week, month, country and chat.
API for GEO starts a step earlier. Its prompt tracking endpoints upsert the prompt list with POST /v1/queries, then run it:
curl -X POST https://api.apiforgeo.com/v1/jobs/execute \
-H "Authorization: Bearer $API_FOR_GEO_KEY" \
-H "Content-Type: application/json" \
-d '{
"models": ["chatgpt", "perplexity", "google-ai-mode"],
"executionsPerQuery": 1
}'
{
"created": 4,
"updated": 0,
"job": {
"id": "job_7c1f9ad2",
"status": "queued",
"models": ["chatgpt", "perplexity", "google-ai-mode"],
"queries": 4,
"creditsCharged": 12,
"createdAt": "2026-09-03T09:41:02Z"
}
}
Four prompts on three models at one run each is twelve credits, and creditsCharged says so in the response. POST /v1/queries/stats reads the result back per prompt as runs, visibility, avgPosition, entities[], sources[] and responses[], and the pricing page states that creating prompts and reading stats are not billed. The practical consequence for an agency: a client who asks on Thursday what five new prompts look like gets an answer on Thursday, for the price of those executions, rather than on the next scheduled collection.
What a ten-client book costs each month
Take ten clients, 30 tracked prompts each, four engines, collected daily. That is 10 × 30 × 4 × 30, or 36,000 AI answers a month, and both vendors bill that same multiplication.
| Route | Monthly bill at 36,000 answers | Effective price per 1,000 | Floor you pay in a quiet month |
|---|---|---|---|
| API for GEO, pay as you go | $468 | $13 | $0 |
| Peec AI agency Scale, 65,000 credits | $795 | $22.08 | $795 |
| Peec AI agency Growth, 25,000 credits | plan too small for 36,000 | $19.80 if filled | $495 |
| Peec AI agency Essential, 10,000 credits | plan too small for 36,000 | $24.50 if filled | $245 |
| Peec AI brand Pro, 150 prompts on 3 models | covers one client, not ten | $18.15 if filled | $245 |
Our arithmetic on published figures: API for GEO's $0.013 per credit, and Peec AI's rule of one credit per prompt, per model, per day. Credit allowances come from peec.ai/pricing-agencies and the plan amounts from peec.ai/ai-instructions, a page dated July 2026. Prices checked on October 23, 2026, pages read again on October 30, 2026.
Two readings of that table are both fair. Filled to the brim, Peec AI's Scale plan is the cheaper answer per unit: 65,000 credits for $795 is $12.23 per 1,000, under the $13 API for GEO charges on its first tier, where its published rate falls to $0.011 above 50,000 credits a month. Half filled, the same plan costs $24.46 per 1,000 and the pay-as-you-go route still costs $13.
Three levers change the arithmetic before you sign anything:
- Tracking frequency. Weekly tracking consumes one third of the daily credits, and Peec AI publishes it on Scale and Comprehensive only. The same ten-client book drops from 36,000 to 12,000 credits a month, which fits Growth at $495.
- Pitch work. Peec AI's agency tiers include pitch projects outside the client quota, three on Essential, ten on Growth, twenty-five on Scale, so prospect research does not eat paid capacity. On a usage API, a pitch costs whatever the pitch runs.
- Churn. A client who leaves in March takes their credits with them on a usage API and leaves the plan fee intact on a subscription. Budget the floor, since the floor is the cost of your smallest month.
What comes back in the payload
| Field an agency report needs | API for GEO | Peec AI |
|---|---|---|
| Answer text | answer_text, responses[] on the stats endpoint | messages[] with role and content on GET /chats/{chat_id}/content |
| Cited URLs | sources[] with url, domain, category, mentionsBrand | sources[] with url, domain, citation_count, citation_position |
| Links written into the answer | not published separately | links[] with url, domain, text, link_count, link_position, added October 27, 2026 |
| Brand mentions | entities[], aliases collapsed onto one entity | brands_mentioned[] with id, name, position |
| Share of voice | shareOfVoice on the model comparison endpoint | share_of_voice metric, formula published |
| Position | avgPosition | position metric and per-brand position in the chat |
| Sentiment | sentiment 0 to 100, themes, per-brand perception | sentiment metric 0 to 100, plus brand perception attributes, objections and fact-checking verdicts |
| Fan-out questions | /v1/fan-outs/top, /overview, /by-query/{queryId} | List Fanout Search Queries and List Fanout Shopping Queries |
| Shopping | /v1/shopping/products with avgPrice, priceRange, avgRating, topMerchants | products, shopping summary, trend, performance, demand, merchants |
| Ads and maps in the answer | not published | ads[] with ad unit type and cards, maps[] |
| Your own site | not published | crawl insights, AI referrals, agent readiness, URL content versions and diffs |
Field names published in each vendor's API reference, checked on October 23, 2026 and read again on October 30, 2026.
Peec AI wins the breadth contest on published surface, and we think an agency should say so out loud rather than pretend the comparison is one-sided. Ad units inside AI answers, map results, content diffs on tracked URLs and agent-readiness audits have no counterpart in the other payload, and the metric formulas being printed in the documentation makes a client question easy to answer. The model channel concept is the detail we would copy elsewhere: a channel is "a stable identifier for a specific AI surface such as ChatGPT UI, Perplexity API, or Google AI Overview that stays constant as the underlying model is upgraded", so filtering by model_channel_id keeps a year-long series intact while filtering by model_id fragments it at every model swap.
API for GEO's side of that table is narrower and arrives in one call per answer rather than per report, with theme clustering and per-brand perception computed for one credit, and reads over stored runs unmetered. Its honest weaknesses: $13 per 1,000 is roughly ten times what a scraping API charges for raw answers, the key arrives on request rather than at sign-up, and four of the ten engines plus the webhooks, OpenAPI spec and SDKs come from our October 26 test rather than from the vendor's own pages. To try it, request a key on apiforgeo.com: it comes with a sandbox project.
Resale, in the words of the two contracts
This is the clause that decides whether you sell a report or a product, and the two answers differ.
Peec AI's terms of use say at 1.3 that "Peec is not appointing Customer as a reseller of the Services", and at 1.6 that "Customer has no right to distribute or allow access to the stand-alone APIs". Section 1.5(a) bars providing access to, distributing, selling or sublicensing the services to a third party, with an exception written into 1.3(b) for delivering the service to an End Customer during your engagement, after which the End Customer needs its own agreement with Peec. Alongside that, the agency page is explicit about what you can do: Looker Studio dashboards cloned per client with read-only links that open without a Peec login, CSV and API feeds into BigQuery, Tableau or Power BI, and an MCP-driven reporting loop that posts a weekly summary per client.
API for GEO's terms read the other way at section 6: "The output is yours to use, store, transform and resell. Reselling is an intended use: put the numbers in your own product, under your own brand, and charge for them." Section 5 draws the one line that remains, forbidding you to sublicense direct API access as distinct from reselling the output. That is the clause behind its 8.8 in our ranking of AI visibility APIs for agencies, which keeps only vendors whose terms allow resale, and 9.0 in the general ranking of 55 APIs.
Four traps on either route
- Filtering a Peec AI report by brand before aggregation. The documentation on filtering and dimensions states that "filtering by
brand_idinfiltersalways reportsshare_of_voiceas 1.0. That number is meaningless." Usehaving, which filters after aggregation and keeps the denominator. An agency that gets this wrong ships a report saying every client owns 100% of the conversation. - Summing tag-level rows by hand. A chat that carries two tags counts under both, so adding the tag rows double-counts. Request the figure without the tag dimension when you need a total.
- Comparing a plan fee to a rate per answer. $795 a month and $13 per 1,000 only compare once you write down your monthly answer count. Do that first, then read the table above.
- Changing the prompt list, the engine list or the country on the switch date. Those three are the denominator of every metric here, and moving one makes the step in the chart your own doing. Run both for two weeks and note the break date under the graph.
Our view
For an agency building a client-facing product on the data, API for GEO is our pick: resale named as an intended use, all ten engines, $13 per 1,000 answers with no floor in a month where two clients pause, and the ability to run a prompt the hour a client asks. The trade-off we would state to a client is the documentation gap, since four engines and the tooling come from our own test rather than from the vendor's pages.
For an agency whose deliverable is a report rather than a product, Peec AI is a defensible buy and the cheaper one per answer at volume: $12.23 per 1,000 on a filled Scale plan, nine of our ten engines plus six surfaces outside the grid, published metric formulas, Looker Studio dashboards clients open without a login, and pitch projects that keep prospecting free. You are buying a platform with an API on it, and the terms keep the data inside that relationship.
Two things to settle before either contract. Ask for the tracking frequency and the project count in writing, because those two numbers, not the sticker price, set your cost per client. And read the resale clause against what you actually plan to ship. Our guide to Profound alternatives with a self-serve API covers the wider field at the same prices, and our explainer on query fan-out covers the array both products return under different names.
Frequently asked questions
Does Peec AI have an API, and which plan includes it?
Yes. Its API introduction, read on October 30, 2026, states that access is available on Pro, Advanced, Enterprise and agency plans, and that Starter plans do not include it. The change is dated: the API changelog records it under v0.39.0 on October 27, 2026, where the same documentation limited the Customer API to Enterprise customers before that. The cheapest brand plan carrying a key is Pro at $245 a month, for 150 prompts, three models and three countries.
How much does Peec AI cost per 1,000 AI answers?
Between $12.23 and $24.50 per 1,000 on its agency plans, using the published credit rule of one credit per prompt, per model, per day. Essential at $245 buys 10,000 credits, which is $24.50 per 1,000; Growth at $495 buys 25,000, or $19.80; Scale at $795 buys 65,000, or $12.23. Those are rates for a filled plan, so an agency that burns half its credits pays double per answer. Credit allowances are published on the agency pricing page and the amounts on peec.ai/ai-instructions, a page dated July 2026.
Can an agency resell Peec AI data, or API for GEO data?
API for GEO's terms say the output is yours to use, store, transform and resell, and that putting the numbers in your own product under your own brand is an intended use; section 5 stops short of letting you sublicense direct API access. Peec AI's terms of use state at 1.3 that Peec is not appointing the customer as a reseller, and at 1.6 that the customer has no right to distribute or allow access to the stand-alone APIs, while 1.3(b) does allow you to deliver the service to an End Customer during your engagement. In practice Peec AI covers a client report under your logo and API for GEO also covers a product you sell.
Which AI engines does each one cover?
API for GEO covers all ten engines in our grid: ChatGPT, Perplexity, Google AI Mode, Claude, Copilot and Grok are listed on apiforgeo.com, and Gemini, Google AI Overviews, DeepSeek and Meta AI were confirmed in our hands-on test on October 26, 2026. Peec AI covers nine of the ten, with Meta AI missing, and goes wider than our grid with OpenAI Search API, Qwen, Mistral, Muse Spark, Naver AI and Amazon Rufus. Peec AI tracks ChatGPT, AI Overviews, AI Mode, Copilot, Perplexity and Gemini on every plan and sells the rest as add-on models.
Can you run a prompt on demand through Peec AI's API?
No endpoint in its reference executes a prompt. You create prompts, brands, tags and topics through the API, and the platform collects answers on the tracking frequency your plan pays for, daily by default and weekly on the Scale and Comprehensive agency plans. API for GEO works the other way round: POST /v1/jobs/execute takes a model list and an executionsPerQuery count, charges one credit per prompt, per model, per run, and returns a job to poll. Pick on that difference if your reporting runs on a schedule you control.
What rate limits do the two APIs publish?
Peec AI documents 200 requests a minute per project, with X-RateLimit-Limit, X-RateLimit-Remaining and X-RateLimit-Reset headers and a 429 when you pass it. API for GEO publishes 300 requests a minute and 10 concurrent executions on its pay-as-you-go tier, with custom limits on committed volume. Both are reading ceilings rather than collection ceilings, so size a nightly job against them before you promise a client a morning report.