Perplexity

AI-powered answer engine

chatno codeFree tier

Overview

Perplexity combines AI with real-time web search to provide sourced answers. Every response includes citations you can verify. Pro ($20/mo) unlocks unlimited Pro searches with GPT-4, Claude, and more. Great for research and fact-checking.

Best For

+Research
+Fact-checking
+Current events
+Academic work

News & Updates

Perplexity AI Loses Bid to Dismiss Reddit Lawsuit Over Data Scraping

Perplexity AI, the AI-powered search engine, has lost its bid to dismiss a lawsuit filed by Reddit over alleged unauthorized data scraping. Reddit accused Perplexity of using its content to train AI models without permission, violating its terms of service and copyright. A federal judge ruled that the case can proceed, allowing Reddit to pursue claims that Perplexity scraped and republished user posts. This decision sets a precedent for how AI companies may be held liable for using platform data without explicit consent.

new_tools8/1/2026

Perplexity Launches Model Council to Compare 8 AI Assistants on Business Questions

Perplexity has launched 'tokenmaxxing Model Council', a feature that runs up to 8 AI models in the cloud to provide multiple perspectives on ambiguous business issues. The feature is designed to help users weigh different viewpoints before making decisions, with the cost being surprisingly affordable. This move positions Perplexity as a multi-model aggregator, competing with single-model assistants like ChatGPT and Claude. The feature is part of Perplexity's broader push into enterprise and professional use cases, offering a unique approach to AI-assisted decision-making.

breakthroughs7/28/2026

Perplexity launches AI agent for Windows PCs that accesses local files and apps

Perplexity has expanded its agentic Personal Computer tool to Windows, allowing computers running the world's most popular OS to be used as a locally run AI system. Like the Mac version that Perplexity launched in April, Personal Computer for Windows operates like a "general-purpose digital worker" that can access local files and apps to perform actions on your behalf, such as creating documents and updating…

new_tools7/28/2026

Researchers Propose LLM Memory Management Method That Outperforms SnapKV by Treating Token Eviction as Estimation Problem

Researchers Maruthi Vemula and Neeraj Praneeth Gajula propose a new framework for LLM memory management, treating eviction decisions as a fixed-lag smoothing estimation problem rather than a heuristic accumulation. Their method, detailed in a new arxiv paper, reframes the choice of which tokens to keep in bounded working memory as estimating a hidden signal, using past and future context. This approach outperforms existing methods like StreamingLLM, H2O, and SnapKV on long-context benchmarks, reducing perplexity by up to 15% while maintaining similar memory budgets. The work suggests that measuring the relevance of past tokens via estimation is more effective than heuristic accumulation rules.