PluelyPluelyDocs

Features

Choosing Your AI: CLI, Your Providers, or Pluely Models

Pluely lets you decide how it gets answers. The model picker in the overlay offers three groups, and you can switch any time.

Where to select a model

  • In the overlay — click the model name on the right side of the composer toolbar (bottom of the overlay, in both Ask and Listen). The picker opens with a search box on top; start typing to filter, click a model to switch instantly.
  • In the dashboard — the Home page has a "Select pro models" button that sets your default Pluely model, with each entry showing whether it handles text only or text + images.
  • CLI and your own providers are set up once in Dashboard → Dev space (Connect your CLI / Provider Selection); after that they appear in the overlay picker under Your CLI and Your providers (Dev Space).

Choosing the best model

  • The Pluely models list is sorted strongest-first, so when in doubt, pick from the top.
  • Need current information (prices, news, anything recent)? Pick Web Search or Perplexity Search, pinned at the very top.
  • Attaching screenshots or images? Pick a model marked as vision-capable (text + image) — image features like Capture and "Use image" need one.
  • Long meetings or heavy documents favor the stronger models; quick lookups run fine (and fast) on lighter ones further down the list.

1. Your CLI

Run completions through an AI CLI already installed on your computer — for example Claude Code, the Gemini CLI, OpenAI Codex, Qwen Code, or Ollama for fully local models. When a CLI is connected and enabled, it takes priority over everything else, and nothing leaves your machine beyond what that tool itself sends.

Best for: developers who already live in a terminal, anyone who wants local/offline models via Ollama, or teams with strict data policies.

2. Your Providers (Dev Space)

Bring your own provider API key. You define a provider once — its endpoint and how to authenticate — and Pluely sends requests directly to it using your key. Your key stays on your device in an encrypted store.

Best for: people who already pay for an AI provider and want to use their own credits, or who need a specific model.

3. Pluely Models

Use Pluely's managed models — no keys, no setup. Your request goes to Pluely, which runs the model for you and streams the answer back. This is the simplest option and is included with paid plans (subject to your plan's monthly limits).

Best for: anyone who wants it to just work without configuring providers.

Web search models

At the top of the Pluely models list sit two special entries — Web Search and Perplexity Search. Pick one when the answer needs current information from the internet (prices, news, docs, sports scores): the model searches the web as part of answering and cites what it found. They're always pinned first so they're one tap away mid-conversation.

Ordered by capability

The rest of the Pluely models list is sorted by model pricing from most capable to lightest, so the strongest models are always at the top. A search box in the picker filters the list instantly — start typing a model's name to jump to it. The exact same catalog and order your plan includes is what you see; nothing is hidden.

How Pluely decides

When you send a question, Pluely picks the path in this order:

  1. Your CLI, if a CLI is connected and enabled — it always wins.
  2. Pluely Models, if you're on a plan that includes managed AI.
  3. Your Providers, using the provider and key you configured.

Privacy note

With Your CLI and Your Providers, requests go to the tool or provider you chose, on your credentials. With Pluely Models, Pluely handles the request for you and never exposes the underlying infrastructure to the app. Whichever you pick, error messages are always friendly and never leak technical internals.

Last updated 2026-07-10 · pluely.com