Trustpilot reviews (1.6/5 average) report widespread billing issues. Users charged twice within 12 days for Enterprise Max ($811 + $725). Cancellation requests ignored, cards charged after cancellation. Refunds rarely honored despite policies claiming otherwise.

Users report Perplexity cut Deep Research usage by 97% after marketing 'unlimited deep research' to Pro subscribers. Unannounced changes to plan limits and false advertising regarding file uploads. Paying users felt deceived.

Terms state subscription payments are non-refundable with no partial credits. Refund window is unreasonably short. When billing issues occur, customers have little recourse. Company discretion on refunds rarely exercised in favor of users.

Enterprise Max costs $325/seat/month ($3,250/year annually). For large teams, costs escalate rapidly. Despite premium pricing, support and features don't match expectations. Some question value vs competitors.

Enterprise Max customers ($3,250/year) report being redirected to chatbot 'Sam' with no human response for 4+ days. Support contradicts promises of 'highest level of support' for premium tiers. No clear way to speak to a real person.

In November 2025, Reddit posts revealed Perplexity began downgrading chats to inferior models without warning. Users tracked the shift and found degraded response quality. Company blamed 'engineering bug' but trust was damaged.

Despite citing sources, Perplexity misquotes, misattributes, and takes content out of context. Average user encounters AI-generated sources within 3 queries. Some citations link to homepages rather than actual content. 45% hallucination rate in some tests.

Perplexity flagged for 'citing real sources with fabricated claims.' Users report finding terms claimed by Perplexity couldn't be found in cited sources. The appearance of rigor misleads when connection between answer and source is tenuous.

Hacker News users report a switch from Sonar to GPT-5 was experienced as a 'massive intellectual downgrade.' Quality changes without user consent or notification. Platform seems optimized for speed over accuracy.

The New York Times sued in December 2025 for 'large-scale unlawful copying.' Reddit sued in October 2025 for 'industrial-scale' scraping. Additional lawsuits from Britannica, Merriam-Webster, Nikkei, and Asahi Shimbun. Legal uncertainty for users citing Perplexity content.

The NYT lawsuit alleges Perplexity intentionally ignored Robots Exclusion Protocol and circumvented 'hard-blocks.' This raises concerns about the company's approach to data ethics and could affect content availability if publishers crack down further.

G2 reviews note the platform cannot find all sources and sometimes links to homepage of sources rather than exact URLs. This undermines the core value proposition of cited, verifiable answers. Research workflows suffer.

G2 reviewers complain that Spaces members cannot add Notes or Comments in chats. Users want collaborative thread features but the platform doesn't support them. For team research, this is a significant limitation.

Perplexity is retrieval-first, designed for research not creation. For creative writing, coding, or ideation, ChatGPT and Claude significantly outperform. Users expecting versatile AI assistant are disappointed.

G2 reviewers note some find responses occasionally slow or generic, especially on complex queries. The $20/month cost feels steep when quality doesn't match expectations. Performance varies based on query complexity.