To track brand mentions in ChatGPT from your own code, use Scavio's ChatGPT API. POST /api/v1/chatgpt/prompt sends the question your buyer would ask and returns ChatGPT's answer as Markdown, every page it cited with publisher and date, the web searches it ran to answer, and any products it recommended. Count your brand in the answer, your domain in the sources and your competitors in both. A prompt costs 2 credits, about $4 per 1,000 prompts on the Growth plan billed yearly, and failed prompts are not charged.
You need a third-party API because OpenAI does not offer one for this. The OpenAI API returns a completion from the model and tools you configure in your own request; it is not the answer ChatGPT gives the people asking it, with the sources ChatGPT chose. There is no OpenAI dashboard of brand mentions either.
We build Scavio, so read this with that bias in mind. Every Scavio number below comes from live calls we made on October 27, 2026. Every competitor detail was read off the vendor's own page or documentation that day, and where a price was not published we say so. The screenshots are each vendor's own page, captured the same day.
Which ChatGPT brand mention API fits your job?
| Tool | Choose it for | What one prompt returns | Assistants | Price (checked Oct 27, 2026) |
|---|---|---|---|---|
| Scavio | Your own tracker, report or agent, with the raw answer behind every number | Answer, cited pages with publisher and date, ChatGPT's search queries, product cards | ChatGPT, Gemini, Google AI Mode | About $4 per 1,000 ChatGPT prompts (2 credits, Growth yearly); 50 free credits |
| DataForSEO | Large prompt batches that can wait in a queue, or a pre-collected mentions database | Answer blocks, cited sources and search results; mention counts in LLM Mentions | ChatGPT, Gemini | $1.20 per 1,000 (queued), $4.00 per 1,000 live; Mentions $0.10 a request + $0.001 a row; $50 minimum top-up |
| Bright Data | Batches delivered as records | Prompt, answer text and HTML, citations, links, recommendations | ChatGPT, AI Mode, Gemini, Perplexity, Copilot, Grok | $1.50 per 1,000 records; 5,000 free a month |
| Apify | Scheduled no-code runs to a dataset | Answer text and sources | ChatGPT Search | From $3.00 per 1,000 searches |
| Ahrefs Brand Radar | Share of voice across Ahrefs' prompt index plus your own prompts | Mentions, impressions, share of voice, cited pages and domains | ChatGPT, AI Overviews, AI Mode, Gemini, Perplexity, Copilot, Claude, Grok | API from Lite ($129/mo); AI add-on from $199/mo; custom prompts from $50/mo for 2,500 checks |
| Peec AI | A managed dashboard your team logs into, with API read access | Visibility, share of voice, sources and chats for tracked prompts | 3 models per plan | Pro $245/mo for 150 prompts (API from Pro); Starter $95/mo |
ChatGPT names the same few brands, and not always the same ones twice
A brand mention API is only as useful as what you do with it, so we ran a test in a category we do not sell into. We sent 15 CRM buying prompts, from "What is the best CRM for a small business?" to "Best CRM for nonprofits", through Scavio's ChatGPT endpoint twice each, and through Google AI Mode twice each for comparison: 60 answers. We counted which CRM brands each answer named and whether the brand's own site was among the cited pages.
ChatGPT named HubSpot in 27 of the 30 answers, Zoho in 19, Pipedrive in 18 and Salesforce in 14. HubSpot was missing only from both nonprofit answers and one of the two real-estate answers. Three things in the data matter for anyone building a tracker:
- One check is not a measurement. 8 of the 15 prompts returned a different set of brands on the second run. ChatGPT kept 71 of 88 brand mentions from one run to the next (81%); Google AI Mode kept 53 of 93 (57%). Salesforce appeared in only one of the two AI Mode runs for 5 prompts.
- ChatGPT picks the shortlist before it searches. Every one of the 30 answers came with
search_queries, and in every answer at least one of those searches already named brands, such as "best CRM for small business 2026 HubSpot Zoho Pipedrive Salesforce pricing features". 118 of the 159 brand mentions in the answers were brands already in ChatGPT's own search query. - A mention and a citation are different numbers. ChatGPT cited the named brand's own site for 154 of its 159 brand mentions: a hubspot.com page all 27 times it named HubSpot, and Pipedrive 18 of 18. Google AI Mode did so for 21 of 146: it also named HubSpot 27 times but cited hubspot.com in only 4, and never cited pipedrive.com. ChatGPT averaged 21.8 cited pages per answer; AI Mode averaged 4.2, and 8 of its 30 answers cited none.
If your report says "we appear in ChatGPT", it should say how often, on which prompts, across how many runs, and whether the answer linked to you. That needs the raw answer, the sources and the searches, which is what the first API below returns.
1. Scavio: the answer, its citations and the searches behind it
Scavio's ChatGPT API is one POST endpoint. Measured on October 27 across the 30 CRM answers above:
- Answer:
answer.markdownis the full answer with ChatGPT's inline citation links, so brand order and the context of each mention are there to parse. Every answer ran on thegpt-6model that ChatGPT picked by default, andmodelreports it. - Sources: 14 to 29 cited pages per answer, each with URL, title and publisher, and a publish date where ChatGPT shows one. TechRadar was cited in 14 of the 30 answers.
- Searches:
search_queries[]held 1 to 3 searches per answer, the queries your content has to rank for to be read. - Products:
shopping[]lists product cards in order on consumer prompts; it was empty for every CRM prompt. - Speed and cost: a median of 20 seconds per answer, between 12 and 59 seconds, at 2 credits. A failed prompt is not charged, and nothing is cached, so every call is a fresh answer.
The Gemini endpoint takes the same prompt and returns the same response shape for 5 credits, and Google AI Mode costs 1 credit, so one parser covers all three assistants. What it does not do: there is no dashboard or share-of-voice chart, so you write the counting, and it does not cover Perplexity, Copilot or Claude. Per prompt, DataForSEO's queued mode and Bright Data are cheaper. The benefit is one key for the answer and for what the answer cites: AI Mode cited YouTube and Reddit in 14 of its 30 CRM answers each, and the same key reads those threads and videos through Scavio's Reddit and YouTube APIs. How to track brand mentions in ChatGPT answers walks through the code.
2. DataForSEO: queued ChatGPT answers and a mentions database
DataForSEO's ChatGPT Scraper API returns ChatGPT answers as typed Markdown blocks with parsed tables, cited sources and the underlying web search results. It has three delivery modes: $0.0012 a result in the standard queue (up to 45 minutes), $0.0024 priority (up to 5 minutes) and $0.004 live (up to 90 seconds), so $1.20 to $4.00 per 1,000. Its LLM Mentions API is different: it queries a database of more than 25 million ChatGPT prompts already collected, for $0.10 a request plus $0.001 a row. Accounts start with $1 of free credit and a $50 minimum top-up. Choose it for large batches that can wait, or for category-wide mention counts on prompts you did not write.
3. Bright Data: ChatGPT answers as records
Bright Data's ChatGPT Scraper API collects the prompt, answer text, answer HTML, citations, attached links, recommendations, country and response URL. It costs $1.50 per 1,000 records pay-as-you-go, or $499 a month for 384,000 records, and every account gets 5,000 free records a month. The same product line covers Google AI Mode, Gemini, Perplexity, Microsoft Copilot and Grok. It does not list ChatGPT's search queries among its fields. Choose it when the deliverable is a file of answers for many prompts and assistants rather than a call inside your app.
4. Apify: scheduled ChatGPT Search runs
ChatGPT Search Scraper, maintained by Apify, had 659 users and a 3.7 rating on October 27. It returns the query, ChatGPT's answer text and the cited sources with title and URL, from $3.00 per 1,000 searches, with failed queries written to a separate errors dataset. Actors run as jobs that write to a dataset, which suits weekly exports into a spreadsheet. Other Apify actors bundle ChatGPT with Perplexity and other assistants for citation monitoring; each is maintained by its author, so pin the one you test.
5. Ahrefs Brand Radar: share of voice across a prompt index
Brand Radar measures mentions, impressions and share of voice across Ahrefs' own index of AI prompts, which it puts at 27,976,130 ChatGPT prompts a month, and across ChatGPT, AI Overviews, AI Mode, Gemini, Perplexity, Copilot, Claude and Grok. Its API has endpoints for mentions, mention history, AI responses and cited pages and domains. Requests that return only your custom-prompt data are free; requests on Ahrefs' prompt data use API units. Pricing lists API and MCP access from the Lite plan at $129 a month, the AI add-on from $199 a month, and custom prompt packs from $50 a month for 2,500 checks ($0.02 a check after that). Choose it for category-level share of voice you could not afford to collect yourself.
6. Peec AI: a tracking dashboard with an API
Peec AI runs the prompts you track daily and reports visibility, share of voice and sources in a dashboard with unlimited users. Plans are $95 a month for 50 prompts (Starter), $245 for 150 (Pro) and $495 for 350 (Advanced), each choosing 3 models, with custom Enterprise pricing. Its API docs give programmatic access to the same data on Pro, Advanced, Enterprise and agency plans, not Starter. Choose it when marketing wants a tool to log into and engineering only needs to pull the numbers into a report.
Profound sells AI visibility tracking with API access on its Enterprise plan, whose price is not published; its pricing page offers a free 7-day trial of 50 prompts. Otterly lists API access on its Standard plan ($189 a month, 100 prompts, 2,000 API requests a month) and Premium plan ($489, 5,000 requests). Semrush's AI Visibility Toolkit offers exports and scheduled reports; we found no documented public API endpoint for its AI visibility data on October 27.
Pick by the number you need
If you want the share of ChatGPT answers that name you and your competitors, the work is running the same prompts often enough for the rate to settle, and keeping the answer, sources and searches so you can see why it moved. Tracking 100 buyer prompts, twice each, every day is 6,000 ChatGPT prompts and 12,000 credits a month: about $24 at the Growth-yearly rate of $2 per 1,000 credits. Running the same 100 prompts twice each, twice a week, is about 1,730 prompts and 3,500 credits a month, which fits the $30 Project plan's 7,000 credits with room to add Gemini or AI Mode. For comparison, Peec AI's Pro plan tracks 150 prompts for $245 a month. Try the ChatGPT API free, 50 credits, no card, or run a prompt in the playground.
Reference: ChatGPT API overview · ChatGPT API docs · Gemini API · Google AI Mode API · free AI visibility checker
Frequently asked questions
Is there an API to track brand mentions in ChatGPT?
OpenAI does not publish one: its API returns completions from the model and tools you configure, not the answer ChatGPT gives its users. Third-party APIs fill the gap. Scavio's ChatGPT API returns the answer, every cited page and the searches ChatGPT ran for 2 credits a prompt; DataForSEO, Bright Data and Apify sell ChatGPT answers per request or record; Ahrefs Brand Radar, Peec AI, Profound and Otterly sell tracking dashboards with API access on paid plans.
How do you measure brand visibility in ChatGPT?
Pick the prompts your buyers ask, run each one several times, and count the share of answers that name your brand and the share that cite your site. On October 27, 2026, 15 CRM buying prompts run twice each through Scavio's ChatGPT API named HubSpot in 27 of 30 answers and Pipedrive in 18, which is the kind of rate worth reporting.
Are ChatGPT answers the same every time?
No. Across the 15 CRM prompts we ran twice, 8 prompts returned a different set of brands the second time, and ChatGPT kept 71 of 88 brand mentions (81%). Google AI Mode varied more: it kept 53 of 93 (57%). Treat any single check as a sample, not a result.
How do you rank in ChatGPT answers?
ChatGPT searches the web before it answers commercial questions, and in our CRM test every one of the 30 answers ran at least one search with brand names already in the query. Being on the review and comparison pages it reads matters: TechRadar was cited in 14 of the 30 answers. The brand's own pricing and product pages were cited almost every time the brand was named.
What is the cheapest ChatGPT brand mention tracking API?
Per prompt, DataForSEO's queued ChatGPT scraper at $1.20 per 1,000 results and Bright Data at $1.50 per 1,000 records. Apify's ChatGPT Search Scraper starts at $3.00 per 1,000 searches, and Scavio and DataForSEO's live mode both work out to about $4 per 1,000 prompts. Dashboard tools price by tracked prompts: Peec AI's Pro plan is $245 a month for 150 prompts.