Perplexica vs Perplexity: which to use, and the open-source system you can own

What Perplexica and Perplexity each are

Perplexica is an open-source AI answering engine you run yourself. It searches the web through SearXNG, reads the top results, and writes an answer with cited sources, all on hardware you control. The maintainer now publishes it as Vane at github.com/ItzCrazyKns/Vane, and the old Perplexica repository address redirects there, so the name people search for and the repository they reach are two words for one project.

Perplexity is a hosted AI answer engine. It selects among several top models for each query, answers with inline citations, and runs entirely in Perplexity's cloud. Its plans run from a free tier to paid Pro and Max tiers, and the model set it names includes ChatGPT, Google Gemini, Anthropic Claude and NVIDIA Nemotron, listed on Perplexity's pricing page.

Where each one runs, and what it costs

Self-hosting Perplexica means running the app and a search backend on a machine you own. Its README recommends Docker, and the image ships with the SearXNG search engine inside it, so one command starts the whole thing. Running Perplexica costs nothing beyond your own hardware and any model API you connect, and the project lists no hosted plan of its own.

Perplexity runs in Perplexity's cloud and sells plans by the month. The free tier carries limited daily usage and basic models, Pro is $17 per month billed annually, and Max is $167 per month billed annually, as Perplexity's pricing page states. There is no self-hosted Perplexity to stand up.

Self-hosting buys control and privacy, and costs a machine and some setup. Hosting buys zero setup, and costs control over where the queries and the answers live.

Models and whose keys they use

Perplexica supports local models through Ollama and connects to OpenAI, Anthropic Claude, Google Gemini and Groq, so a private search can stay on your machine while a harder question borrows a cloud model. You supply the keys. A local model and a cloud model both return the same cited-answer shape, and the difference is where your question travels.

Perplexity chooses the model itself and lets Pro and Max accounts set a preferred one. You do not bring your own key, and the model choice is part of the subscription.

Which one fits

A person who wants private answers and will run a stack should install Perplexica, now Vane. A person who wants cited answers in seconds and would rather not run anything should pay for Perplexity. Both are answer engines, and neither is built to run the work itself.

When an answer is not enough

Kortix is the open-source AI Operating System: your agents, their skills, your company memory, and every connector in one git repo you own, with the agents working on real cloud computers. Where Perplexica and Perplexity return an answer, Kortix runs the work and keeps the context.

• Each session has its own isolated machine and branch.
• Session work reaches main through a change request. Merge is default-deny for agents.
• Any model provider with your own keys.
• Run it on Kortix Cloud, in your VPC, or on your own on-prem network. Self-host is free.

For a team that wants to own the whole system, Try Kortix.