Claude vs Perplexity

One Perplexity event accounts for most of the distance between these two. On 2026-02-06 Perplexity silently cut Pro Deep Research from around 500 queries/day to roughly 20/month — a ~95% collapse dressed as 'allocating more compute per session' — while the same quiet stroke halved Labs to 25/month and capped formerly unlimited uploads at 50/week, drawing complaints that an annual plan had been 'quietly gutted.' Claude has its own unpopular moves: two weekly caps on 2025-08-28 and, notably, silently accelerated 5-hour depletion during weekday mornings on 2026-03-23. What differs is the sequel. Claude reversed that peak throttling on 2026-05-06 and doubled Pro/Max 5-hour limits alongside it, and even administrative shifts (the 2026-06-11 move to fixed weekly resets) were documented; its one severe disruption, the government-forced Fable 5 / Mythos 5 suspension on 2026-06-12, was external, model-specific, and undone by July 1 rather than a deliberate quota cut. Perplexity does offset with real loosening — the $200 Max tier's uncapped research from 2025-07-17, Comet going free worldwide — but its core Pro value dropped an order of magnitude without notice. For a stability-minded commitment Claude is the more trustworthy: numeric caps, explicit reversals. Perplexity only makes sense if you live inside its reduced Pro allowances or pay for Max, where research is uncapped and the silent-cut risk that defines its Pro history stops biting.