Gemini vs Perplexity: cited answers from Perplexity, a 5 TB bundle from Gemini, both at $20

Which is better, Gemini or Perplexity?

The 72% comes from a study the European Broadcasting Union and the BBC published on 21 October 2025. Journalists at 22 public broadcasters in 18 countries rated more than 3,000 answers to news questions from the free versions of ChatGPT, Copilot, Gemini and Perplexity. Sourcing was the widest gap in the whole study. It wasn't close.

Read the date before the number. The answers were collected in late May and early June 2025, and the free Gemini app the BBC tested ran Gemini 2.5 Flash, per the EBU report. Google has shipped Gemini 3 since, and nobody has rerun the test on the paid plans. What hasn't changed is the design underneath, which the next section quotes from Google's own pages: Gemini searches when its model decides to, and Perplexity searches every time. That design is why we still read the split below as current.

Does Perplexity cite its sources better than Gemini?

Yes. Perplexity cites every answer by design, and Gemini cites when its model decides a search will help. Perplexity's pricing page lists "Citations in every answer" on the free plan. Its help center says Pro brings "10x as many citations per answer". Google says the opposite about Gemini, in plain words, on its sources help page: "Not all responses include related links or sources." No Sources button means no links were given.

Google's developer docs explain why. On Grounding with Google Search, Google writes that "The model analyzes the prompt and determines if a Google Search can improve the answer." When it decides it can't, you get an answer from the model alone. That's fine for a recipe. It's a problem for a figure you'll repeat in a meeting.

The EBU and BBC study shows what that looks like across hundreds of answers. 42% of Gemini's responses had no direct source at all, meaning no URL pointing to a specific article. In 54% of them, evaluators found phrases like "sources such as RTBF say" with a link to a different outlet or no link at all, and no other assistant passed 4% on that measure. One Gemini answer for NPR said it drew "from NPR sources" while none of its 11 links were from NPR.

Perplexity fails the other way: too many links. Asked by Norway's NRK how many people died in the Myanmar earthquake, it listed 19 URLs and referred to three of them in the answer. A wall of citations looks checked. It isn't, so click the link behind any number you plan to use. For how Perplexity compares with a writing assistant rather than a search assistant, see Perplexity vs ChatGPT.

Gemini Deep Research vs Perplexity Deep Research: which can you run more?

Gemini lets a free account try Deep Research, and Perplexity gives a free account exactly 1 Research query a month. That's the only Deep Research count either company prints for its consumer plans. Google lists Deep Research on the free tier of its plans page. Its limits page adds a warning for people without a plan: "some features that require more compute resources, like Deep Research, may be unavailable during periods of high demand".

On the paid plans neither one counts. Google AI Pro gets "4x higher than standard limits", and Google adds that Deep Research "will consume more of your usage" from that shared pool. Perplexity Pro gets "Monthly limits (average use)" for Research, per its plan comparison last modified 2 October 2026. The only paid Perplexity tier with a printed number is Enterprise Pro, from $40 a seat: 50 Research queries a month.

What each one reads is the bigger difference. Gemini Deep Research can pull from Gmail, Google Drive and Google Chat as well as Search, a change 9to5Google reported on 5 November 2025. If your project notes already sit in Docs, that saves an upload step. Perplexity Pro searches "across the web, files, and apps" and adds paid data from PitchBook and Statista, so it's the stronger tool for a market question your inbox can't answer. One hands-on test cuts the other way, though. In G2's April 2026 comparison, Gemini won the deep research task, while Perplexity won summarization and academic paper analysis. That's one reviewer's prompts, not a benchmark. Still, a cited report isn't automatically a thorough one.

Google AI Pro vs Perplexity Pro: what does each $20 plan include?

Google AI Pro is a storage plan with AI attached, and Perplexity Pro is a research tool with other labs' models attached. The prices match within a cent. (Google also sells AI Plus at $4.99 and AI Ultra from $99.99 a month, and Perplexity's consumer step up is Max, at $200.) What comes with the two $20 plans doesn't match, as Google's US plan page listed it on 11 October 2026 and Perplexity's pricing and help pages listed it between August and early October 2026:

Gemini Notebook is the product Google called NotebookLM until 16 July 2026, and it's the closest thing Google sells to Perplexity's research workflow: you load your sources and ask questions of them. Perplexity has no storage to sell. If you'd buy Google storage anyway, is Google AI Pro worth it does the math line by line.

Both browsers now do the same job. Comet, Perplexity's browser, has been free for everyone since 2 October 2025, but its assistant queries sit behind Pro: the plan table gives Free "No" on that row. Google puts Chrome auto browse on AI Pro. Pick the browser you already use.

How many prompts do you get on each? Neither prints a number

Neither Google nor Perplexity publishes a prompt count for its $20 plan as of October 2026. Google describes Gemini's limit as "compute-based" and says it factors in "the complexity of your prompt, the models and features you use, and the length of your chat". Then comes the one rule it does commit to: "Your limit refreshes every 5 hours until you reach your weekly limit." You can see where you stand under Settings, then Usage Limits.

Perplexity uses words instead of numbers too. Pro Search is "Weekly limits (average use)", and free accounts get 3 Pro Searches a day. Enterprise Pro prints 400 a week, which is the nearest yardstick. Perplexity Pro limits goes through each pool and what the open source trackers assume.

Each one has a fallback when you hit the wall. Google tells paying users they "can continue your conversation with Flash-Lite". A Perplexity Pro user can switch to Best mode, which runs without a quota. Neither fallback is the model you chose. If Gemini stops you mid project, Gemini usage limit reached sorts out which limit it was.

Can you use Gemini inside Perplexity?

Yes, but on Pro it's Gemini Flash, not Gemini Pro. Perplexity's model table, a snapshot dated 4 September 2026, lists Gemini 3.7 Flash among its eight Pro search models. Its older "What is Perplexity Pro" page, last modified 3 September 2026, still names Gemini 3.1 Pro. Perplexity says the model selector in your account is "always the source of truth", so check there before you pay for one model.

It's also not quite the same Gemini. Perplexity says it runs third party models with "its own search, citations, tools, prompts, safety systems, and usage limits", so a model there "can behave differently" from the same model in its maker's app. That's the trade. You get Gemini answers with Perplexity's citations on top, and you lose Google's 1 million token window and its Gmail and Drive access. The Gemini app, for its part, runs Google's models and nobody else's.

Which one trains on your chats by default?

Both do, and both let you turn it off. Google's activity help page says "If you're 18 or over, Keep Activity is on by default." With it on, chats are kept for 18 months unless you change the period, and Google's privacy hub says "A subset of chats are reviewed by human reviewers". Reviewed chats are kept for up to three years, even after you delete your activity.

Turning Keep Activity off stops training on future chats, but Google still keeps them for 72 hours. Temporary chats work the same way and "are not used to train Google's AI models." That's the cleaner switch for a one off sensitive question.

Perplexity's data collection page, last modified 3 September 2026, says "For Free, Perplexity Pro and Perplexity Max users, AI Data Retention is enabled by default." The toggle sits in Account settings, under Preferences. The catch is timing: "Previously collected training data cannot be deleted or removed." So switch it off before your first real question, not after. Enterprise Pro is the one Perplexity plan whose data is never used for training.

Can one subscription cover Gemini and Perplexity?

Not from either company. Google AI Pro doesn't include Perplexity, and Perplexity Pro's only Gemini is Flash. Whizi Pro, at $29.99 a month for 2,500 credits, carries Perplexity's Sonar Pro at 20 credits an answer and Gemini 3.1 Pro at 10 in one picker, with web search on by default. Side by side compare is on Pro too, so one question goes to both and the answers sit next to each other (comparing models side by side). That's 125 Sonar Pro answers or 250 Gemini Pro answers a month. It's not a replacement for either bundle: there's no 5 TB, no Gemini Deep Research over your Gmail, no Comet, and Perplexity's own Deep Research model isn't in Whizi's catalogue. If those are why you're buying, buy the original (Gemini in Whizi lists what is there).