In 20 seconds
- As of October 23, 2026, ChatGPT costs $0, $8 (Go), $20 (Plus) or $100–$500 (Pro) a month. Perplexity lists Free, Pro at $17 a month billed annually and Max at $167 a month billed annually; its help center prices Max at $200 a month or $2,000 a year.
- Both use OpenAI's newest models: Perplexity's changelog added GPT-6 Astra on September 21 and GPT-6.1 Sol and Claude Opus 5.5 on October 5, 2026. ChatGPT is ad-supported on Free and Go; Perplexity dropped its ad test, the FT reported in February 2026.
- Both train on consumer chats by default unless you opt out: ChatGPT's 'Improve the model for everyone' and Perplexity's 'AI data retention' both start on, per each company's help center. In the Tow Center's March 2025 study, Perplexity got 37% of citation queries wrong and ChatGPT search 134 of 200.
Contents
ChatGPT vs Perplexity comes down to what you want from an AI assistant as of October 23, 2026. OpenAI's ChatGPT is a general assistant with search built in, plus Work and Codex modes and always-on agents on its top plans. Perplexity started as an answer engine that cites sources in every reply, and also sells Perplexity Computer, an agent that works across your files and apps. Both charge about $20 a month for the main paid tier and $200 for the top one, and Perplexity's search runs on many of the same models OpenAI sells.
This comparison is based on OpenAI's and Perplexity's pricing pages, help centers, release notes and changelogs, checked October 23, 2026, plus a citation study by Columbia's Tow Center and a news-integrity study by the European Broadcasting Union. Prices are US list prices before tax. For how ChatGPT stacks up against Claude and Gemini, see our three-way comparison.
The short version
- Price: ChatGPT Plus is $20 a month; Perplexity Pro is $17 a month billed annually, and Perplexity's pricing page shows no monthly-billed Pro price. At the top, Perplexity Max is $200 a month and ChatGPT Pro has tiers at $100, $200 and $500.
- Cheapest paid plan: ChatGPT Go at $8. Perplexity has nothing between Free and Pro.
- Models: Perplexity lets Pro users choose models, including OpenAI's GPT-6 Astra and GPT-6.1 Sol and Anthropic's Claude Opus 5.5, according to its changelog. ChatGPT runs only OpenAI models.
- Agents: ChatGPT has Work mode on every plan (limited on Free and Go) and dots on Pro. Perplexity Computer is included from Pro, with credits.
- Ads: ChatGPT shows ads on Free and Go. Perplexity ended its ad test, according to the FT.
- Training: both say their consumer training setting starts on; you can opt out on either.
How much do ChatGPT and Perplexity cost?
| Price point | ChatGPT | Perplexity |
|---|---|---|
| Free | Free: GPT-6 Luna in Chat; may show ads | Free: "basic AI models," citations in every answer, limited uploads |
| Under $10 | Go, $8: Free's models with higher limits; may show ads | — |
| About $20 | Plus, $20 a month | Pro, $17 a month billed annually; monthly price not shown |
| About $100 | Pro 100, $100 | — |
| About $200 | Pro 200, $200 | Max, $200 a month or $2,000 a year ($167 a month) |
| $500 | Pro 500, $500 | — |
| Teams | Business, $25 per user monthly or $20 billed annually | Enterprise: prices not shown on the pricing page |
Table 1Both companies' main paid plan sits near $20 and their top individual plan at $200, but only ChatGPT sells an $8 plan and a $500 one. Source: learn.chatgpt.com/docs/pricing, perplexity.ai/hub/pricing and Perplexity's Max help article, checked October 23, 2026.
Last verified: October 23, 2026 (full rows with sources in 2026-10-23/data/P31-plans-compared.csv)
ChatGPT Pro is billed monthly only, per OpenAI. Perplexity's help center says Max's annual billing is available only on the web. For every ChatGPT tier in detail, including dots and Ultrafast, see ChatGPT plans compared.
Which models do ChatGPT and Perplexity use?
| ChatGPT | Perplexity | |
|---|---|---|
| Free | GPT-6 Luna in the Chat tab | "Basic AI models" (not named) |
| About $20 | Chat: GPT-6 Sol. Work and Codex: GPT-6.1 Sol and GPT-6 Luna, limited GPT-6 Astra | "Access to top AI models" with model selection; changelog adds GPT-6 Astra (Sep 21) and GPT-6.1 Sol and Claude Opus 5.5 (Oct 5) |
| Top tier | Pro: the Pro reasoning option in Chat runs on GPT-6 Astra; Pro 500 adds GPT-6 Astra and GPT-6.1 Sol Ultrafast | Max: "Frontier AI models" and "advanced reasoning models" |
Table 2ChatGPT runs only OpenAI models; Perplexity offers a choice that includes OpenAI's and Anthropic's latest. Source: OpenAI's pricing page, Pro tiers help article and October 7 release notes; Perplexity's pricing page and changelog, checked October 23, 2026.
Last verified: October 23, 2026
ChatGPT runs different models in different places. Since October 7–8, the Chat tab has used GPT-6 Sol on paid plans and GPT-6 Luna on Free and Go, per OpenAI's release notes; GPT-6.1 Sol is "available in Work and Codex, not Chat," per OpenAI's pricing page.
Perplexity's own pages disagree with each other. Its changelog lists GPT-6 Astra, GPT-6.1 Sol and Claude Opus 5.5 as new models in September and October 2026, but its "What is Perplexity Pro?" help article, checked the same day, still lists GPT-5.2, Claude Sonnet 4.6 and Gemini 3.1 Pro. The help article also says access to advanced models "may be limited during weeks of especially heavy usage." Its pricing page is inconsistent too: the Free card lists "basic AI models," while its FAQ says Free includes "access to the top AI models with limited weekly usage."
How do search and citations compare?
Search is the core of Perplexity and one feature among many in ChatGPT.
- Perplexity shows "citations in every answer," including on Free, per its pricing page. Pro answers carry "10x as many citations per answer," according to its help center, and Pro adds premium data sources such as PitchBook and Statista.
- ChatGPT search is available on every plan and to people who aren't signed in, per OpenAI's help center; logged-out search began February 5, 2025, per OpenAI's search announcement. Responses that use web search "may include citations," and a Sources button lists them when available. OpenAI's help article says ChatGPT search "sometimes partners with other search providers" and links Microsoft's privacy statement for how those providers handle queries.
Both offer a slower deep-research mode that writes a cited report. OpenAI says "Deep research usage varies by plan" and shows remaining tasks in a usage counter; it doesn't publish numbers. Perplexity lists "Deep research" on Pro and "Expert level research" on Max, also without numbers.
If you build on search rather than use an app, the trade-offs differ: our AI web search API comparison covers Perplexity's Search API and OpenAI's web search tool at API prices.
Which has the better agent?
They take different approaches.
- ChatGPT Work runs longer tasks and builds deliverables. It's limited on Free and Go and included in full from Plus, sharing one allowance with Codex. Dots, OpenAI's always-on agents running on GPT-6 Astra, come with Pro (for adults outside the EEA, UK and Switzerland), Business Premium and Enterprise, where an admin has to turn them on; our dots explainer covers what they can do.
- Perplexity Computer is included from Pro, with "4,000 bonus credits," and Max adds "10,000 monthly credits" plus 35,000 bonus credits, according to Perplexity's pricing page. Free has no Computer access. On Mac, Computer can run fully on the device for Pro, Max and Enterprise subscribers, as our Perplexity Computer story explains, and its setup asks you to install Perplexity's Comet browser.
Neither company publishes how many tasks an allowance or credit balance buys.
File uploads
| ChatGPT | Perplexity | |
|---|---|---|
| Size per file | Documents and presentations 512 MB; spreadsheets about 50 MB; images 20 MB; text files capped at 2M tokens | 50 MB "for all file types" |
| Upload counts | Free: 3 a day; up to 80 files every 3 hours | Free "limited"; Pro "10x" |
| Storage | 25 GB per user | Not stated for Free, Pro or Max; Enterprise Pro and Max files kept 7 days |
Table 3ChatGPT accepts documents ten times larger than Perplexity's 50 MB cap and publishes a storage cap; Perplexity states a retention period only for Enterprise uploads. Source: OpenAI's file-uploads article; Perplexity's file-uploads, pricing and data-collection help articles, checked October 23, 2026.
Last verified: October 23, 2026
Ads, privacy and training
| ChatGPT (Free, Go, Plus, Pro) | Perplexity (Free, Pro, Max) | |
|---|---|---|
| Ads | Free and Go only; Free can switch to "Ads-Free" with lower limits | None; ad test ended (FT, via Search Engine Land, February 18, 2026) |
| Training on chats | Data sharing "is enabled by default" on Free, Plus and Pro; turn off Improve the model for everyone | "AI data retention" is "enabled by default"; turn it off in Preferences |
| Opting out | Applies going forward | Applies going forward; "Previously collected training data cannot be deleted" |
| Private mode | Temporary Chat, kept up to 30 days | Not covered in the data-collection article |
| Business plans | Business, Enterprise and Edu not used for training | "Perplexity Enterprise data is never used for AI training purposes" |
Table 4Neither consumer service is opt-in for training; both help centers say the setting starts on. Source: OpenAI's ads and data-controls help articles; Perplexity's data-collection help article; Search Engine Land's summary of the FT, checked October 23, 2026.
Last verified: October 23, 2026
OpenAI says ads don't appear in Temporary Chats or for accounts identified as under 18; our ads explainer covers the controls. Perplexity's ad decision is reported, not announced: the FT, as summarized by Search Engine Land, quoted unnamed executives saying even labeled ads could undermine trust in its answers.
What do independent evaluations say?
No published head-to-head covers the October 2026 versions. The two best-known studies are from 2025 and tested older models, so they are background, not a ranking of the October 2026 products.
| Study (date) | What it measured | ChatGPT | Perplexity | Limits |
|---|---|---|---|---|
| Columbia Journalism Review / Tow Center (March 6, 2025) | 200 news excerpts per tool; could the tool name the right article, publisher and URL? | ChatGPT search: 134 of 200 answered incorrectly | Free: 37% incorrect; Pro answered more correctly but had a higher error rate than Free | One query per excerpt; eight tools; not meant to be "extrapolated" to other models |
| EBU and BBC (October 22, 2025) | Journalists rated 3,000+ news answers from ChatGPT, Copilot, Gemini and Perplexity | Per-assistant rate not in the release | Per-assistant rate not in the release | 45% of all answers had a significant issue; Gemini had issues in 76%, more than double the others |
Table 5In the Tow Center's test, Perplexity's free version got fewer citations wrong than ChatGPT search, but both tested versions have since been replaced. Source: CJR (March 6, 2025) and EBU (October 22, 2025).
Last verified: October 23, 2026
The Tow Center found paid tools, including Perplexity Pro, gave "definitive, but wrong, answers" instead of declining. A 2026 experiment by a Peec AI researcher shows how easily AI search tools can be fed false statistics; our analysis of that case covers what it does and doesn't prove.
Which should you pick?
| If you mainly… | Consider | Why |
|---|---|---|
| Want a free assistant without ads | Perplexity Free | ChatGPT Free and Go may show ads; Perplexity ended its ad test, per the FT |
| Want the cheapest paid plan | ChatGPT Go ($8) | Perplexity has no plan between Free and Pro |
| Research topics and want a source on every answer | Perplexity Pro | Citations in every answer; 10x the citations on Pro; premium data sources |
| Want to switch between OpenAI and Anthropic models | Perplexity Pro | Model selection, with GPT-6.1 Sol and Claude Opus 5.5 per its changelog |
| Write, analyze large files or code | ChatGPT Plus ($20) | 512 MB documents, Codex and Work included |
| Want an agent that works on your Mac's own files | Perplexity Pro or Max | Computer runs locally on supported Macs |
| Want always-on agents that keep working between chats | ChatGPT Pro (from $100) | Dots are on Pro, Business Premium and Enterprise only |
| Want the least training on your chats without a business plan | Either, after opting out | Both train by default unless you turn it off |
| Buy for a team | ChatGPT Business or Perplexity Enterprise | ChatGPT publishes $25/$20 seats; Perplexity's prices come from sales |
Table 6Options by need, based on the prices, features and settings above; neither service fits every row.
What's confirmed and what may change
- Confirmed (official pages, October 23, 2026): ChatGPT's prices and models; Perplexity's Free, Pro (annual) and Max prices; Perplexity's changelog model additions; file limits; ad eligibility on ChatGPT; both companies' training settings.
- Reported: Perplexity's end of advertising (FT, via Search Engine Land, February 18, 2026).
- Uncertain or not published: Perplexity Pro's monthly-billed price and Enterprise seat prices; which models each Perplexity plan gets as of October 23, 2026, since its help article and changelog disagree; deep-research limits on both services; how many Computer tasks a credit buys.
About this story
Based on the sources linked below. How we report: Editorial standards.
Sources · 19
- ChatGPT Work and Codex pricing — OpenAI
- ChatGPT release notes — OpenAI Help Center
- Introducing ChatGPT search (with updates) — OpenAI
- About ChatGPT Pro tiers — OpenAI Help Center
- Meet dots — OpenAI
- Searching the web with ChatGPT — OpenAI Help Center
Show all 19 sources
- Deep research in ChatGPT — OpenAI Help Center
- Uploading files and audio to ChatGPT — OpenAI Help Center
- Ads in ChatGPT — OpenAI Help Center
- Data controls in ChatGPT — OpenAI Help Center
- Pricing — Perplexity
- Perplexity Max — Perplexity Help Center
- What is Perplexity Pro? — Perplexity Help Center
- Changelog — Perplexity
- File uploads — Perplexity Help Center
- Data collection at Perplexity — Perplexity Help Center
- AI Search Has a Citation Problem — Columbia Journalism Review, Tow Center (March 6, 2025)
- AI's systemic distortion of news is consistent across languages and territories — EBU (October 22, 2025)
- Perplexity stops testing advertising — Search Engine Land, summarizing the Financial Times (February 18, 2026)


Comments (0)