Yes, Perplexity remembers you between conversations. Its Memory feature draws on saved memories (preferences, interests and details you share) and your past searches. You can turn either one off in Settings, under Memory, delete saved memories there, and delete past searches from your History. Max subscribers using Perplexity Computer also get Brain, a separate memory with its own controls.

Everything below is checked against Perplexity's help center on October 16, 2026. Perplexity changes these settings often, so the linked pages are the source of truth.

Does Perplexity remember previous conversations?

Yes, in two ways (Perplexity Help Center):

Memories: the preferences, interests and information you share. Perplexity stores key details as you ask questions, building a memory bank you can manage in Settings.
Search history: your past questions and answers, used when they make a new answer more relevant.

Perplexity decides when to use them based on the question, and when a memory shapes an answer, it shows up among the answer's sources. Memory works across all of Perplexity's models and search modes (Perplexity changelog, Dec 2025), so switching models inside Perplexity doesn't reset it.

It does stay inside Perplexity. Perplexity doesn't import chat history from ChatGPT or Claude (Perplexity Help Center), and what it remembers is managed in Perplexity's own settings.

What is thread continuity in Perplexity?

Thread continuity means a conversation keeps the context of your earlier questions, so follow-ups build on previous answers without you repeating the background. Perplexity now calls threads sessions. A session holds your first question, every follow-up, every answer and the sources used, and it's designed to keep context for follow-up questions. If you switch topics often within one session, it may need some guidance to stay on track (what is a session, session capabilities).

Continuity within a session and Memory across sessions are different features. A session remembers one conversation; Memory carries details from one conversation into the next.

What is Perplexity Brain?

Brain is a separate, self-improving memory for Perplexity Computer. It builds a working model of your projects, people and files from your sessions, connected tools, the files you create in Computer and your corrections, and it runs in the background (what is Brain). Each entry links back to its source, and the entries are organized into Concepts, Entities and Workstreams that you can browse and edit.

As of October 2026, Brain is rolling out in research preview to Max and Enterprise Max subscribers using Computer. Search doesn't access Brain, and Brain ignores incognito sessions. Perplexity's own summary: Memory helps Computer know you, and Brain gives Computer the working context to move your work forward.

Where are Perplexity's memory settings?

Go to Settings → Memory (Perplexity Help Center). You'll see:

Use search history: a toggle for referencing past searches.
Notes: a toggle for saving simple details from your sessions for future answers.
Manage memories: your full memory bank, where you can search, filter and delete entries.
Brain (Max subscribers): a toggle for building structured memories.

If you remember Memory living somewhere else, it moved: the sidebar item that was called Customize is now Plugins, and Memory has moved to Settings. Plugins is rolling out gradually, so some accounts still show Customize for now (Perplexity Help Center).

How to delete Perplexity memory

To delete one memory or all of them:

Open Settings → Memory.
Click Manage memories to open your memory bank. Search or filter for the entry you want.
Click the trash icon next to a memory to delete it, or use clear all to delete every memory at once.

A deleted memory leaves the memory bank and isn't used in future answers. Perplexity may keep a log of cleared memories for up to 30 days for safety, debugging and to make sure memories aren't recreated right after deletion (Perplexity Help Center).

To stop a past conversation from shaping future answers, delete that session from your history. Open History from the sidebar; you can delete individual sessions or all of them (what is a session).

To stop Perplexity from using memory going forward, turn off Notes and Use search history in Settings → Memory. Each toggle works on its own. Turning Memory off doesn't clear existing memories, so delete them in Manage memories as well.

To edit or delete what Brain has learned, open the Memory page in Computer, where every entry can be edited or deleted. To stop Brain from writing new knowledge, turn it off under Settings → Memory settings → Brain (what is Brain).

For a one-off private question, use incognito. Memory and previous searches are always off in incognito, and questions asked there are never retained or used in memory.

What does Perplexity memory keep, and what doesn't it?

Perplexity says it has taken steps to reduce how much sensitive information, like health or financial details, memory retains. It also says memory is generated by a model that favors how often you ask about something and what you share, so check your memory bank if you've discussed sensitive topics (Perplexity Help Center).

On consumer accounts, Perplexity may use content from memory to improve its models. Its Memory article says you can turn that off with AI Data Retention in Settings → Preferences → Artificial Intelligence, though its Brain and Enterprise articles describe that setting as covering searches only.

How is Enterprise memory different?

Enterprise memory works the same way for each user, with differences in control and ownership (Memory for Enterprise organizations):

Admins control Reference search library (Use search history) and Reference saved memories (Notes) for the whole organization. If an admin turns one off, it's off for everyone.
Memories created in an Enterprise organization belong to the organization. A user who leaves loses access to them.
When admins disable Memory for the organization, existing memories are deleted, and the same applies when a member disables it for their own account.
Perplexity never uses memory content to train models for Enterprise users.

Projects add another layer. In Computer, a project can have a shared Brain that everyone in the project sees, and the project owner decides whether sessions also draw on the personal memory of whoever started them. The help center's two articles describe the project controls slightly differently, so check the one that matches your account type.

Check what Perplexity remembers with a test

Use a harmless, made-up detail, like a demo project's code name. Note where you shared it, whether it shows up in Manage memories, and what a fresh conversation cites when you ask about it.

Then change the detail and ask a related question. If the old answer persists, check whether it comes from a saved memory, a past session or text already in the new conversation. Each route needs a different fix: delete the memory, delete the session, or start a new one.

Building memory like this into your own app

Perplexity's memory is a product feature, so it only works inside Perplexity. If you're building an assistant and want the same behavior, the design questions carry over: can users inspect what's remembered, see where it came from, correct it and delete it, and predict when it will be used?

A memory API gives you those parts without building them from scratch. With supermemory, each user gets a namespace, their profile comes back with one call, and most profile entries carry a memory ID you can forget (profile API). The user-preference guide walks through it with code, the long-term memory guide covers the design, and the MCP context-sharing guide shows how to share one memory across assistants like Claude and ChatGPT through our MCP server.

How Perplexity memory works: what it remembers and how to delete it