Use Perplexity Models in Big-AGI.

Running Perplexity in Big-AGI

Add your Perplexity API key and run the Sonar models at Perplexity's own rates. Big-AGI isn't a reseller: no markup and no intermediary, so the bill comes straight from Perplexity to your account.

• Your key, your billing. Usage is billed by Perplexity directly, nothing routes through us.
• Web search, with the dials exposed. Every Sonar model grounds its answers in live web results; Big-AGI surfaces the controls Perplexity's own app keeps hidden: search context size, a publication-date filter from unfiltered up to one year back, and an academic-only mode that restricts sources to peer-reviewed research.
• Real reasoning over search results. Sonar Reasoning Pro runs DeepSeek R1's chain-of-thought over what it finds; Sonar Deep Research goes further, running many searches at a configurable reasoning effort to produce a full report instead of one grounded answer.

Why not just Perplexity?

Perplexity's app gives you one search engine in one box. Big-AGI runs a Sonar model next to Claude, GPT, or Gemini in the same conversation, so you can cross-check a live, cited web answer against a frontier model's own reasoning instead of trusting either one alone.

Perplexity Pro runs $20 a month for unlimited use in its own app. The API bills differently: per token like everything else, plus a per-search fee that scales with how much web context a query pulls in. A few grounded lookups a day can land cheaper than the subscription; a Deep Research report running dozens of searches usually won't. Beam parallel runs add more searches on top, the honest tradeoff for comparing models side by side.

Your keys and your data

Turn on Direct Connection and the browser calls Perplexity directly, bypassing the Big-AGI server, when your key is client-side and Perplexity's API allows it. Your keys stay in your browser. Chats are stored locally first, and sync only if you turn it on. The AI Inspector shows the exact request, the token counts, and a cost estimate.

Perplexity in Beam

Put Sonar into a Beam next to Claude, GPT, and Gemini, so a live, cited web answer sits right beside frontier-model reasoning. Fusions then combine, cross-check, and synthesize the parallel answers instead of just picking the best one. Parallel runs mean more tokens, and for Perplexity, more searches too.