OpenAI
GPT-6 and GPT-5.6 families
-
=GPT()gpt-6.1-terra, default -
=GPT6()gpt-6-astra, most capable -
=GPT61()gpt-6.1-sol -
=GPT6mini()gpt-6-luna, cheapest -
=GPT56()gpt-5.6-sol -
=OpenAI(p, s, model)any OpenAI model
Updated October 2026: GPT-6, Gemini 4 and the Perplexity Agent API
Type =GPT(), =SCRAPE() or =SERP() in any cell.
GPTSheetsPlus brings the latest OpenAI, Gemini and Perplexity models, Bright Data web
scraping and Exa web search into Google Sheets. Free trial, then $19 once. Bring your own
API keys.
No subscription. No account. Keys stay inside your Google account.
| A | B | C | D | |
|---|---|---|---|---|
| 1 | Product | Website | Scraped content (=SCRAPE) | Price/mo (=EXTRACT_INFO) |
| 2 | Notion | notion.so/pricing | # Pricing · Plus $10 per member / month… | 10 |
| 3 | Linear | linear.app/pricing | # Plans · Basic $8 per user / month… | 8 |
| 4 | Figma | figma.com/pricing | # Figma plans · Professional $16 per seat… | 16 |
| 5 | Slack | slack.com/pricing | # Pricing · Pro $8.75 per user / month… | 8.75 |
| 6 | ||||
| 7 | Question | =PPLX (web-grounded answer) | =GPT (one-line summary) | |
| 8 | Latest Notion AI news | Notion launched AI Agents with workspace-wide… | Notion shipped autonomous Agents… | |
| 9 | Who is Linear's CEO? | Karri Saarinen is the co-founder and CEO of Linear [1] | Karri Saarinen | |
AI models
Every function takes a prompt and an optional system prompt. Retired model names are mapped to their replacements automatically, so sheets you built last year keep working.
GPT-6 and GPT-5.6 families
=GPT() gpt-6.1-terra, default =GPT6() gpt-6-astra, most capable =GPT61() gpt-6.1-sol =GPT6mini() gpt-6-luna, cheapest =GPT56() gpt-5.6-sol =OpenAI(p, s, model) any OpenAI model Gemini 4, 3.8 Flash and 3.1 Pro
=GEMINI() gemini-3.8-flash, default =Gemini4() gemini-4-argon, flagship =Gemini31Pro() gemini-3.1-pro =Gemini38Flash() gemini-3.8-flash =Gemini35FlashLite() fastest, cheapest =GEMINI(p, s, model) any Gemini model Agent API with live web search
=PPLX() fast answer with citations =PPLX_PRO() light multi-step research =PPLX_REASONING() multi-hop research =PPLX_AGENT(p, s, effort) fast to xhigh =EXTRACT_INFO() utility, uses GPT =SUMMARIZE_BULLETS() utility, uses GPT Workflow
Search, scrape, extract, enrich. Every step is a cell, so you can drag it down a column and let the sheet do the work.
=SERP("CRM startups", 5) Exa web search returns URLs =SCRAPE(A2) Bright Data fetches the page =EXTRACT_INFO(B2, "founder") GPT pulls out one field =PPLX("What does " & A2 & " do?") Perplexity adds live context | A | B | C | D | |
|---|---|---|---|---|
| 1 | URL (=SERP) | Page content (=SCRAPE) | Founder (=EXTRACT_INFO) | What they do (=PPLX) |
| 2 | attio.com | # The CRM for the AI era. Attio is built by… | Nicolas Sharp | AI-native CRM for go-to-market teams [1] |
| 3 | folk.app | # folk · The CRM people actually enjoy… | Simo Lemhandez | Lightweight CRM for agencies and founders [1] |
| 4 | clay.com | # Clay · Go-to-market data enrichment… | Kareem Amin | Data enrichment and outbound automation [1][2] |
| 5 | close.com | # Close · The CRM built for growth… | Steli Efti | Sales CRM with built-in calling and email [1] |
| 6 | ||||
| 7 | =SUMMARIZE_BULLETS(B2) | • AI-native CRM • Flexible data model • Built for GTM teams | ||
Web scraping
Powered by Bright Data Web Unlocker, which handles blocks, CAPTCHAs and JavaScript-heavy pages. Results are cached for 6 hours, so recalculating a sheet never burns extra credits.
=SCRAPE(url) page as markdown =SCRAPE_HTML(url) raw HTML =SCRAPE_TITLE(url) page title =SCRAPE_DESCRIPTION(url) meta description =SCRAPE_KEYWORDS(url) meta keywords =SCRAPE_LANGUAGE(url) page language =SCRAPE_LINKS(url) every link on the page =SCRAPE_STATUS(url) HTTP status code =BRIGHTDATA(url, out, render, country) full control =Firecrawl(url) with your Firecrawl key Web search
Exa-powered search that returns a list straight into your sheet. Ask for URLs, root domains or full page text, and filter by category such as company, news or people.
=SERP(query, n, category) returns up to 100 URLs=SERP_DOMAIN() returns clean root domains=SERP_TEXT_RESULT() returns page text, ready for =EXTRACT_INFO()=SERP_COMPANY() searches company websites only| A | B | |
|---|---|---|
| 1 | Result | Status |
| 2 | https://attio.com | 200 |
| 3 | https://folk.app | 200 |
| 4 | https://clay.com | 200 |
| 5 | https://close.com | 200 |
| 6 | https://pipedrive.com | 200 |
| 7 |
Pricing
Try everything for free. Then one payment unlocks lifetime use with your own API keys.
$0no card needed
$19once, forever
Secure checkout via Lemon Squeezy. Your key arrives by email.
FAQ
During the free trial, no. You get 10 free uses each of the OpenAI, Gemini, Perplexity, scraping and search functions on our keys. After that you activate the $19 lifetime license and paste your own keys from OpenAI, Google AI Studio, Perplexity, Bright Data, Exa or Firecrawl. You pay those providers directly, at cost, with no markup.
Yes. $19 once, for lifetime use of the add-on, including future model updates. There is no subscription and no account to create. The only ongoing cost is what you pay the AI and scraping providers for your own usage.
The license and your API keys are stored against your Google account, so they work in every spreadsheet you open. Activate the license from a spreadsheet you own.
Responses are cached by prompt so that recalculating a sheet does not burn credits or trial uses. Pass TRUE as the ignore_cache argument, for example =GPT("prompt", "", TRUE), to force a fresh call. Scrapes are cached for 6 hours.
No. Retired function and model names are mapped to their current replacements automatically. =GPT5() now runs gpt-5.6-sol, =GPT4o() runs gpt-6-luna, and so on. See the legacy names page in the docs for the full list.
There is no login and we do not collect personal information. Your prompts go to the provider you chose, using your own key once you are licensed. API keys are stored by Google in your account’s add-on properties, not on our servers.
Install in a minute. Your first formulas are free.