Perplexity stopped returning citations on three consecutive searches, so I switched to a different query. Same thing. I refreshed, cleared cache, tried a different browser. Still blank citations, still the same confident-but-unsupported answers.
Turns out it was a Focus mode conflict — I’d accidentally switched to a mode that wasn’t appropriate for my query type. Five-second fix once I knew where to look.
That’s the pattern with most Perplexity problems. They’re not deep bugs — they’re configuration mismatches or quota issues or browser quirks that have quick fixes once you’ve seen them before. Here are the nine that come up most.
Fix 1: Perplexity Giving Wrong or Outdated Answers
Start with what Focus mode you’re in. Look at the icon to the left of the search bar — it shows your current mode. Default is Web, which is broad and pulls from whatever’s indexed. That’s fine for general queries, not great for technical or academic topics.
Switch modes deliberately:
- Academic — restricts sources to scholarly databases. Better for research questions, way less noise.
- Writing — no web retrieval, just the model. Use this when you want synthesis without source dependency.
- YouTube / Reddit / Social — niche but genuinely useful when you want community perspectives over official sources.
- Wolfram Alpha — math, science, unit conversions. Do not use this for text queries.
If you’re in Web mode and getting wrong answers, the issue may also be source quality. Perplexity doesn’t filter for authority by default. A random forum post can appear in citations alongside peer-reviewed research. Check the citations — if the sources themselves are thin, the answer will be too.
For time-sensitive information, use the date filter (the calendar icon on the search bar) to restrict results to the past week or month. Perplexity can sound authoritative about outdated information without flagging that it’s outdated.
Fix 2: Pro Search Quota Exhausted
Pro Search is limited to 300 queries per day on the Pro plan. That sounds like a lot until you’re deep in a research session asking follow-ups.
Here’s the thing people miss: every follow-up question in a thread that uses Pro Search counts as a separate Pro query. A 10-turn research thread can burn through 10 Pro searches. Your daily counter is in the bottom left of the sidebar — check it before you’re surprised.
The practical fix is to be selective. Use Pro Search for initial complex queries where you need deep synthesis. For follow-ups, clarifying questions, and “what does that mean?"-style questions, switch to Standard. Standard search is unlimited and handles conversational follow-ups fine.
Counter resets at midnight UTC. If you hit the wall late in the day, that’s only a few hours wait.
If you’re on the free tier and running into rate limits: the free tier gives you 5 Pro Search queries per day. It’s not much. For anyone using Perplexity as a daily research tool, the $20/month Pro plan is actually one of the more defensible AI subscriptions — the query quality difference is real. Our how to use Perplexity AI guide covers what Pro actually unlocks in practice.
Fix 3: Citations Broken or Returning 404s
Perplexity’s citations are snapshots. The source existed when Perplexity retrieved it. If the page has since moved, been deleted, or is behind a new paywall, you’ll get a 404.
This is not a Perplexity failure — it’s how the web works. But it can be frustrating when you need to verify a claim.
Workarounds:
- Paste the URL into web.archive.org — Wayback Machine often has a cached version.
- Add
cache:in front of the URL in Google search (cache:example.com/article). - In Perplexity itself, click the numbered citation and look at the preview panel — sometimes the full text is already extracted there even if the live URL is dead.
If you’re getting 404s on most citations for a specific query, that’s usually a sign Perplexity sourced from low-quality sites that rotate URLs frequently. Switching to Academic focus mode pulls from more stable sources (arXiv, PubMed, Semantic Scholar) where citations don’t go dead.
Fix 4: Perplexity App Not Loading (Mobile)
The mobile app is more fragile than the web version. Three things cause most app failures:
Clear the cache first. On Android: Settings > Apps > Perplexity > Storage > Clear Cache. On iOS, you have to delete and reinstall to get a clean cache. This fixes most “app opens to a blank screen” or “stuck on loading” issues.
Check for updates. Perplexity ships mobile updates frequently. A version mismatch between your app and the current API can cause silent failures — searches that appear to run but return nothing, or a UI that loads but doesn’t respond. Check your app store.
Test on cellular vs. WiFi. Some corporate and university networks block Perplexity’s API endpoints. If the web version at perplexity.ai works on cellular but the app doesn’t work on WiFi, that’s the network. Not fixable on your end without a VPN.
If the app is crashing on launch: delete, reinstall, and don’t restore from iCloud backup — the cached state can carry forward the same issue that caused the crash.
Fix 5: Focus Mode Not Finding Relevant Sources
You switched to a specific Focus mode and you’re getting worse results than general web search. That’s usually a scope mismatch.
Academic mode is the most common one that trips people up. It’s excellent for published research but useless for current events, product questions, or anything that’s primarily discussed on forums rather than in papers. Switching to Academic to get “better” answers on a question like “what’s wrong with my Notion AI” will just return irrelevant research papers.
Match the mode to the question type:
| Question type | Best focus |
|---|---|
| Research / scientific | Academic |
| Product recommendations | Web (default) |
| Community sentiment | |
| Breaking news | Web + date filter (last day) |
| Technical docs | Web, or just link directly |
| Math / conversions | Wolfram Alpha |
If you’re in the right mode and still getting irrelevant results, the query itself may be too vague. Perplexity handles specific queries better than general ones. “Machine learning” returns broad noise. “How does RLHF work in training LLMs” returns useful synthesis.
Fix 6: Perplexity Slow or Timing Out
Slowness breaks down into two scenarios.
Free tier slowness. The free tier sits behind Pro tier in the processing queue. During peak hours (US mornings), this is noticeable. Not a bug, just capacity allocation. If consistent responsiveness matters for your workflow, Pro addresses this.
Model selection on Pro. Perplexity routes queries through different models depending on query complexity. If you’re using Perplexity Pro with the “Best” model option enabled, you’re getting the highest-quality synthesis but also the highest latency. Switch to the “Balanced” option for everyday queries — the quality difference on routine research is minimal, the speed difference is significant.
Connection issues. If Perplexity is slow on one device but fast on another, the issue is local. Check whether a VPN is adding latency — Perplexity routes through US servers, and a VPN exit node far from those servers adds round-trip time. Also confirm you’re not on a throttled mobile connection.
If queries time out entirely (spinner that runs forever): reload the page, try the query again. If it consistently times out on complex queries, try breaking the query into two simpler ones. Very long input prompts hit processing limits that cause timeouts.
Fix 7: Perplexity Spaces / Collections Not Syncing
Spaces (previously called Collections) let you save searches and maintain context across sessions. When they don’t sync, you lose threads you thought were saved.
The actual save behavior: Perplexity saves thread history automatically, but Spaces require a deliberate save action. If you’re relying on auto-save, check that you’re not in a private/incognito session — those don’t persist anything.
Login state across devices. Spaces sync via your account. If you’re not logged in on a device, Spaces won’t appear. Sounds obvious, but Perplexity can appear to work without being logged in (it uses guest sessions), and that’s when people lose threads they thought were saved.
Slow sync. If a Space updated on one device but hasn’t appeared on another: force-refresh the Spaces sidebar (pull to refresh on mobile, F5 on web). Cross-device sync isn’t instant — usually takes 30-60 seconds. If it hasn’t synced after a few minutes, log out and back in.
Fix 8: API Errors (For Developers)
If you’re building on the Perplexity API and hitting errors, these are the common culprits:
401 Unauthorized. Your API key is invalid, expired, or not being passed correctly. Keys go in the Authorization header as a Bearer token: Authorization: Bearer pplx-xxxx. Regenerate from the API settings page if you’re unsure — there’s no cost to issuing a new key.
429 Too Many Requests. You’ve hit rate limits. Perplexity’s API tiers have different RPM (requests per minute) limits. Implement exponential backoff in your code. Don’t retry immediately on a 429 — that makes it worse.
Context length errors. The llama-3.1-sonar-large-128k-online model has a 127k token context window, but your combined prompt + expected output needs to fit within that. Very long system prompts or pasting in entire documents pushes you toward the limit. Trim system prompts and chunk large documents.
Inconsistent source retrieval. The API can return different citations for the same query on different calls — that’s intentional, not a bug. Perplexity retrieves sources at query time, so real-time changes in indexed content affect results. For reproducible outputs, use temperature: 0 and be aware that source variability is still possible even then.
Fix 9: Login and Account Access Issues
Can’t get into your account or your Pro subscription isn’t showing up.
Password reset gotcha. If you originally signed up via Google or GitHub SSO, you don’t have a Perplexity password to reset. Go to the login page and use the same social login you used to sign up. Trying to reset a password on an SSO account will send a reset email that doesn’t work.
Pro subscription not activating. After a payment, Pro features can take up to 10 minutes to activate. Refresh after waiting. If still missing after 15 minutes, check your email for a failed payment notification — banks sometimes block AI subscription charges from unfamiliar merchants. If payment went through but features are still missing, contact support with your transaction ID.
Regional access. Perplexity is available in most countries, but some features (notably some real-time data integrations) are US-only. If you’re using a VPN and notice features disappearing when you switch exit nodes, the VPN is probably routing you through a region where that feature isn’t available.
When to Contact Perplexity Support
If none of the above works:
- Billing issues (charges with no service, plan that won’t activate) always need human review.
- Account lockouts — if you’ve lost access to both your primary email and your social login, contact support with account creation date and the email you used.
- API production incidents — if your API integration is down and you’re in production, open a support ticket with your request ID and response logs.
Perplexity support is at perplexity.ai/contact. Response times are reasonable for Pro users; expect longer on free tier.
If everything above is working but you’re still not getting great research results from Perplexity, the issue might be how you’re using it rather than how it’s running. Our guide to using Perplexity AI effectively covers the query structures and Focus mode combinations that consistently return better synthesis. And if you’re evaluating whether Perplexity is the right tool for your workflow versus something like ChatGPT or Claude, the Perplexity vs ChatGPT and Claude vs Perplexity comparisons break down the actual use-case differences — they’re different tools built for different jobs.
TechSifted is reader-supported. When you buy through links on this site, we may earn an affiliate commission at no extra cost to you. This doesn’t influence our editorial process — we evaluate tools independently.