Overview
AnythingLLM is an all-in-one AI productivity app that turns any document, resource, or piece of content into context for LLM-powered chat. It supports both desktop and self-hosted deployments and works with dozens of LLM providers — including Perplexity as a first-class cloud provider, giving every workspace access to web-grounded answers with citations.AnythingLLM is built and maintained by Mintplex Labs. It supports per-workspace LLM configuration, so you can use Perplexity in one workspace and a different provider in another. Learn more at anythingllm.com.
Setup
1
Get your API key
Generate a Perplexity API key from the API portal.
Get API Key
Generate your Perplexity API key.
2
Open LLM settings
In AnythingLLM, go to Settings → LLM Preference.
3
Select Perplexity AI
Choose Perplexity AI from the provider list and paste your API key.
4
Choose a model
Once your key is entered, pick a model from the Chat Model Selection dropdown (e.g.,
sonar-pro) and save.System vs. Workspace Models
AnythingLLM lets you scope LLM choice at two levels:- System LLM — The default provider used for every workspace and agent.
- Workspace LLM — Per-workspace override that takes precedence when chatting inside that workspace.
Available Models
AnythingLLM’s Perplexity provider ships a fixed list of Sonar model IDs and sends requests to Sonar Chat Completions (POST /chat/completions). Agent API models such as perplexity/sonar and effort levels are not available in this provider. The list includes sonar-reasoning, which is no longer available, so choose sonar, sonar-pro, or sonar-reasoning-pro.
Sonar Chat Completions has been superseded by the Agent API. To see how existing Sonar requests are handled, read the Sonar to Agent API migration guide.
Use Cases
Pairing Perplexity with AnythingLLM gives you:- Live web answers inside document workflows — your workspace can answer questions about uploaded PDFs and the open web in the same chat.
- Citation-grounded retrieval — Perplexity returns source URLs you can audit alongside your retrieval-augmented context.
- No-RAG fallback for current events — when documents don’t contain the answer, the model can reach for live web context automatically.
Agent Configuration
When using AnythingLLM’s AI agents, you can configure Perplexity as the agent’s LLM in the workspace’s Agent Configuration menu. This is useful for agents that need to chain web research with skills like web browsing, document Q&A, or external API calls.Links & Resources
AnythingLLM Perplexity Docs
Official AnythingLLM setup guide for Perplexity AI.
AnythingLLM Docs
Full AnythingLLM documentation.
Perplexity API Reference
Full Perplexity API reference.
Perplexity Models
Available models and capabilities.
Support
Need help with the integration?- Browse the AnythingLLM documentation
- Review our FAQ