The original version of this comparison ran in March, and its core argument was simple: ChatGPT and Perplexity AI are not competing for the same job. ChatGPT is a general-purpose tool for writing, coding, and reasoning. Perplexity is a research tool that puts a cited source behind every answer. Six months on, that framing still holds, and it is still the most useful way to decide between them.
A Note on How This Refresh Was Done
This refresh did not include a fresh check of either vendor's pricing page. The numbers below come from our product database, which records plans and testing notes from hands-on testing in May 2026. AI assistants restructure plans, rename tiers, and swap default models more often than almost any other software category, so treat every price here as a starting point, not a quote. If a number matters to a purchase decision, check it on the vendor's own page first.
Pricing, as Recorded in Our Product Database
ChatGPT (tested May 2026):
| Plan | Price | What it adds |
|---|---|---|
| Free | $0 | GPT-4o mini, limited GPT-4o messages, basic image generation, web browsing, file uploads |
| Plus | $20/month | Full GPT-4o access, image generation, Advanced Data Analysis, custom GPT creation, voice mode |
| Team | $25/user/month (minimum 2 seats) | Higher limits, admin console, workspace management, data excluded from training by default |
| Enterprise | Custom | SSO via SAML, admin analytics, 128K context window, priority support |
| API | Pay per token | GPT-4o at $2.50 input / $10 output per 1M tokens; GPT-4o mini at $0.15 / $0.60 per 1M tokens |
Perplexity (tested May 2026):
| Plan | Price | What it adds |
|---|---|---|
| Free | $0 | Unlimited quick searches, a limited number of Pro searches per day (5 at time of testing) |
| Pro | $20/month | Unlimited Pro searches, model choice (GPT-4o, Claude, Sonar Large), file uploads, image generation |
| Enterprise Pro | Custom | Team features, SSO, data privacy controls |
One difference from the original post worth flagging: the March version listed ChatGPT Team at $30/user/month. Our product database records it at $25/user/month with a two-seat minimum. Team pricing has historically differed between annual and monthly billing, which may explain the gap, but we have not confirmed which figure is current. Check before you budget for a team rollout.
The headline number did not move. At the individual paid tier, both tools cost $20/month, so the decision still comes down to what you get rather than what you pay.
What the Original Verdict Said
The March post landed on this: use Perplexity to gather and verify information, use ChatGPT to analyze it and produce something from it. The deciding factor was not model quality or price. It was the source citation.
When Perplexity answers a question, it attaches numbered links to the sources it used. You can click each one and see whether the source actually says what the answer claims. ChatGPT can browse the web and sometimes includes links, but citing sources is not its default behavior, and most answers arrive with no trail back to where the information came from.
What Still Holds
The citation gap is still the real differentiator. Our product data still lists always-cited sources as Perplexity's top strength and "tasks requiring always-cited sources" as a case ChatGPT is not ideal for. Nothing in our records suggests that has changed.
Both still hallucinate. This needs repeating in every version of this comparison. Neither tool is accurate by default. Perplexity's citations give you a way to catch errors, not a guarantee against them. Our testing notes are blunt about it: Perplexity's citations are not always from authoritative primary sources, and output needs verification before it goes into a report or a decision. Perplexity can also cite a real page that does not actually support the claim. ChatGPT's mistakes are harder to catch because there is usually no source to check against.
ChatGPT is still the stronger generalist. Writing, coding, debugging, long multi-turn conversations, and analyzing uploaded files all remain ChatGPT's home ground. Perplexity is not built to draft a proposal or iterate on code, and our testing notes point users toward ChatGPT or Claude for those jobs.
The two-tool workflow still makes sense. Research the facts in Perplexity, check the sources that matter, then paste the verified material into ChatGPT to turn it into a comparison memo, an RFP question list, or an email to your team.
Where Each One Still Falls Short
ChatGPT's weak spots, from our product data:
- ▸Hallucinates facts with confidence, and usually without a source to check.
- ▸Real-time web information is limited on the free tier.
- ▸Costs climb for larger teams at $25/user/month.
- ▸Context window limits can still bite on very long or complex tasks.
Perplexity's weak spots, from our product data:
- ▸Not a good choice for creative writing or long-form drafting.
- ▸Heavy research use effectively requires Pro, since the free tier caps Pro searches.
- ▸Weaker than ChatGPT at complex, multi-step reasoning.
- ▸Citations are only as good as the pages they point to, and those are not always primary sources.
Privacy and Data Handling
For anyone putting work material into either tool, the data handling matters as much as the features. Here is what our product database records:
- ▸ChatGPT: Conversations on the free tier are used to train models by default. You can opt out in settings. Team and Enterprise plans exclude data from training by default, and API usage is not used for training. Chat history can be exported and deleted.
- ▸Perplexity: Perplexity states Pro plan queries are not used to train models. Free tier queries may be used for product improvement. Search history can be deleted. Our notes describe its privacy policies as still evolving, which is typical for a younger company.
The practical rule is the same for both: do not paste confidential client data, credentials, or regulated information into a free consumer tier. If your organization needs AI research at work, the Team or Enterprise tier is the minimum sensible place to start, and your security team should review the current data terms directly.
Who Should Pick Which
Pick Perplexity Pro if most of your AI use is looking things up: vendor research, fact-checking, current events, or anything where you need to show where an answer came from. The citations save real time when you have to verify before you act.
Pick ChatGPT Plus if most of your AI use is producing things: drafts, code, analysis of uploaded documents, or long working sessions where context matters. It is the more capable all-rounder.
Run both if your work is split. At $20/month each, the pair costs less than many single SaaS seats and covers the full path from research to finished output.
Stay on the free tiers if you use AI a few times a week. Both free tiers are genuinely useful for occasional work. Perplexity's cap on Pro searches is the limit you will notice first.
The Bottom Line
The March verdict holds. Perplexity is the research tool, ChatGPT is the production tool, and at the same $20/month price the choice depends on which kind of work fills more of your day. What changes is how you should treat the numbers: as a snapshot from our May 2026 testing, not a guarantee. Confirm current plans on chatgpt.com and perplexity.ai before you commit, especially for team or enterprise purchases.
For the full original breakdown, see the March comparison of ChatGPT and Perplexity for research. If you are weighing ChatGPT against other general-purpose assistants rather than a research tool, the ChatGPT vs Claude vs Gemini refresh covers that decision.