How to choose which AI to use in Perplexity and get the most out of it

Perplexity has become a Swiss Army knife of artificial intelligenceA single platform from which you can launch advanced searches, conduct in-depth research, write texts, program, summarize documents, and much more, all supported by various top-tier AI models such as GPT, Gemini, Claude, Grok, Kimi, and Sonar itself. The problem is that, when you're given so many options, it's easy to feel overwhelmed and end up choosing models at random without really knowing what you're doing.

If you have Perplexity Pro or are thinking of using it intensivelyKnowing which AI to use at any given time makes all the difference between getting a mediocre response and truly leveraging the tool. In this guide, you'll see, step by step and with clear examples, how to select the right model for each task, what Best mode means, when it's beneficial to activate the "Thinking" models, how to switch between them within the same conversation, and how all of this fits together with other key features like Focus, Copilot, Spaces, and scheduled tasks.

What makes Perplexity special compared to other chatbots?

One of Perplexity's greatest strengths is that it acts as an all-in-one solution. where AI models from various companies coexist in a single interface. You don't have to jump between ChatGPT, Gemini, Claude, or Grok: Perplexity takes care of integrating them and, by default, automatically chooses the engine it deems most appropriate for your query.

Every time OpenAI, Google, Anthropic, or others launch a new modelPerplexity usually adds it to its catalog quite quickly. That's why it's not uncommon to see names like GPT-5.1, Claude Sonnet 4.5, Gemini 3.0 Pro, or Grok 4.1, in addition to their "Thinking" variants and their own Sonar model. This list is updated over time, so the specific models you see may change, but the basic principles of use remain the same.

It is important to understand that the answers are not a 1:1 copy of the official versions. of those models. Perplexity can apply its own system for searching, rewriting, or citing sources, so the final result may differ slightly from what you would see in the native ChatGPT, Gemini, or Claude apps. Even so, you still benefit from its reasoning, creativity, and understanding of complex contexts.

Furthermore, Perplexity is not just your average chat appIt combines web search, document analysis, summaries, and collaborative features. You can see the sources it uses, save conversations in Spaces, schedule recurring tasks, or even use its Comet browser to integrate AI directly into your daily workflow.

All of this makes choosing the right model even more important.If you know which AI excels at reasoning, which performs best in creative writing, or which engine is ideal for technical research, you can fit each model where it really makes sense in your day-to-day life.

How to manually choose the AI ​​model in Perplexity

Changing the model in Perplexity is done directly from the box where you type the prompt.It's present in both the web version and the mobile and desktop apps. It's not hidden, but it can easily be missed if you don't know where to look.

Look at the bottom right of the text field Where you type your question: you'll see several icons, and to the left of that row, one shaped like a chip (or similar) labeled "AI Model" or with the name of the active model. That's the button you need to switch engines.

When you click that icon, a list of all available models appears. In your account: you'll usually see options like Best, Sonar, various GPTs, Gemini, Claude, Grok, Kimi, and the "Thinking" versions of some of them. The list may vary depending on your plan, usage limits, and new features added to the platform.

One very useful detail is that you can change the model at any time within the same conversationYou can ask an initial question using Claude Sonnet 4.5 to delve deeper into a topic, and then switch to GPT-5.1 Thinking for linear code analysis or Gemini 3.0 Pro for a more creative approach, without needing to open a new chat.

Below the list of models you will usually see information about usage limitsPerplexity indicates how many consultations you have left with the most advanced models or with intensive reasoning variants. This is especially important in the Pro plan, where leading models usually have a daily or monthly limit to prevent abuse.

What types of models will you find at Perplexity

Although the specific names change over timeIn Perplexity you will always find several model families which you should be clear about so you don't go crazy choosing.

Best ModeThis is the default option, which lets Perplexity decide which model to use based on the task. It typically combines speed, quality, and cost; perfect if you don't want to overcomplicate things or if you're just starting out and don't yet fully understand the different motors.

Sonar (Perplexity's own model)It's highly focused on searching, summarizing, and providing answers based on recent sources. It's usually very useful for general inquiries, seeing what's being published on a topic, or requesting clear explanations supported by links.

GPT models (e.g., GPT-5.1 and Thinking variants)They excel at content creation, fluent writing, and great versatility. They are usually a safe bet for writing long texts, reformulating content, programming, or combining reasoning with creativity.

Claude models (such as Claude Sonnet 4.5 and Claude Sonnet 4.5 Thinking)They tend to offer highly analytical and well-structured responses, with good handling of lengthy texts and complex reasoning. They are a solid choice for technical research, reports, analysis of long documents, and detailed explanations.

Gemini (for example, Gemini 3.0 Pro)They typically excel at understanding context, connecting different topics, and tasks that blend data, content, and creativity. If you want to explore ideas for marketing campaigns, content, and trends, they are a top candidate.

Grok (as Grok 4.1 and Grok 4.1 Thinking)It's geared more towards quick interaction and, depending on the settings, can provide responses with a more informal tone. It can be useful for brainstorming, general questions and answers, or exploring alternative approaches to the same topic.

Kimi K2 Thinking and other reasoning modelsThese “Thinking” profiles focus on longer and deeper chains of reasoning, especially useful in complex problems of programming, mathematics, logic or step-by-step analysis.

What does “Thinking” mode mean and when to use it

The “Thinking” variants of models such as GPT-5.1, Claude Sonnet 4.5, Grok 4.1, Kimi K2 or Gemini 3.0 Pro They are designed to dedicate more "mental time" to the problem. This usually translates into longer reasoning processes, more structured responses, and a greater capacity to sustain detailed explanations.

They are not magic models that always provide better answersInstead, they are engines tuned to prioritize the quality of reasoning over speed or cost. In other words, they take longer and consume more resources, which is why Perplexity often limits their use to a specific number of queries.

You'll want to use Thinking when: Are you tackling a complex coding problem, do you want AI to analyze a technical document in depth, do you need a detailed explanation of a complex concept, or do you want to compare nuanced scenarios? For example: debugging a difficult bug, designing a software architecture, or breaking down a dense academic paper.

On the other hand, it's not worth wasting time on Thinking consultations For simple tasks like requesting a short summary, writing a brief email, performing a basic information search, or asking a quick general knowledge question, Best mode, Sonar, or a standard GPT/Gemini/Claude model will be more than sufficient.

A quick rule you can apply is the followingIf the task requires you to stop, think, and perhaps make diagrams or numbered steps on paper, the Thinking variant is probably worthwhile. If you would do it "in one go" without complicating things, the standard model will be more than enough.

Which model to choose depending on the type of task

Many Perplexity Pro users encounter the same questionWith so many options available, should I have a default model or change it all the time? The most practical approach is to combine a main model and then make specific changes for particular tasks.

For help with complex code and debuggingThis is where reasoning models shine. GPT-5.1 Thinking, Claude Sonnet 4.5 Thinking, and Kimi K2 Thinking typically provide good step-by-step explanations, suggest architectural improvements, and detect logical errors. For simpler syntax questions or small functions, a standard GPT-5.1 or Best mode is usually faster and uses fewer resources.

For technical research and in-depth analysisSonar combined with Claude Sonnet 4.5 is a very good pairing. Sonar handles searching for up-to-date sources and citing them, while Claude excels at organizing information into clear blocks, comparisons, and comprehensive summaries. If you need it to break down documentation even further, a Thinking version can refine the process even more.

For creative writing (short stories, screenplays, marketing copy)GPT-5.1 and Gemini 3.0 Pro are usually excellent choices. If you're looking for a more lighthearted tone or crazy campaign ideas, Grok can also add a spark. When you want to check the consistency of a long storyline or polish the style in detail, you might want to upgrade to GPT-5.1 Thinking.

For article and document summariesThe Best or Sonar modes meet this need very well because they combine search, text comprehension, and synthesis. If it's a highly technical paper or a dense report where you don't want to miss any nuances, switching to Claude Sonnet 4.5 or a Thinking variant can give you an extra level of detail.

For general questions and answersHere you can leave Best mode enabled and forget about it, or use Sonar if you want it to explicitly cite web sources. If you find the response lacking in nuance, switch to GPT-5.1 or Claude within the same conversation.

How to use Perplexity step by step and fit the model choice

Beyond choosing a model, it's key to understand how Perplexity is used on a daily basis. to get the most out of it, especially if you're going to use it for work, study, or digital marketing.

1. Enter the platformYou can go to www.perplexity.ai and start asking questions without registering. However, if you want access to features like Copilot, history, Spaces, or the advanced model selection, you should create a free account or sign in with Google.

2. Do your first searchType your question in the text box, which usually says something like "Ask a question...". For example: "Digital marketing trends for ecommerce in 2025". Press Enter and let the AI ​​do its work. You'll get an answer in the form of a summary, usually with bullet points or structured paragraphs.

3. Check the sourcesBelow the response, you'll see links to the pages used. This is one of the major differences compared to a pure chatbot: you can check the source of each statement, validate data, find more information, and decide which sources you find most reliable.

4. Adjust the focus of the responseIf the answer doesn't go where you wanted, you can use the suggested questions that appear afterward or write your own clarification. Something like, "Focus the answer only on fashion e-commerce in Spain" or "Explain it for someone with no marketing knowledge." Perplexity will then revise or expand the answer while respecting the previous context.

5. Use Focus mode to filter the source of the informationFocus lets you specify whether you want the AI ​​to prioritize results from Reddit, YouTube, academic publications, financial documents, or other types of sources. For example, when researching a scientific topic, you can activate academic mode to focus on specialized papers and articles.

6. Activate Copilot mode if you have an accountIn the upper right corner, you'll usually see a button called Copilot. When you activate it, Perplexity not only responds but also asks follow-up questions to refine your needs: ideal for exploring a complex project, developing a campaign strategy, or defining a detailed brief.

7. Change model if you are using the Pro versionWith the paid plan (around $20 per month), you can select models like GPT-4, GPT-5.1, Claude, Gemini, and other newer ones. Click on the model name or the chip icon and choose the one that best suits the task. For example, GPT is usually excellent for content generation, and Claude tends to provide very analytical and thorough answers.

8. Leverage AI to generate useful contentOnce you've established the model, you can issue well-thought-out prompts. Examples: "Create a comparative table of SEO strategies for cosmetics companies in Europe" or "Write an email to launch a new line of vegan products in Spain." The more context you provide, the more relevant the response will be.

9. Use your search historyWith a session logged in, Perplexity saves your chats. You can return to them, retrieve key information, continue iterating, and add new nuances without losing track.

10. Integrate Perplexity into your daily workflowCopy answers that you find useful and import them into Notion, Trello, Google Docs, Canva, or wherever you usually work. You can use AI for briefs, video scripts, emails, market research, presentations, or internal documentation.

Perplexity as a tool for writing articles and SEO content

If you work in content marketing, SEO, or copywritingPerplexity is especially powerful because it combines search with text generation. A typical workflow for writing an SEO article might look something like this.

Step 1 – Generate a good titleAsk the model (for example, GPT-5.1 or Gemini 3.0 Pro) something like: “Suggest 5 effective SEO titles for an article about social media marketing for SMEs.” Indicate the tone (informal, technical, educational) and the keywords you want to include. The AI ​​will take into account trends and writing style.

Step 2 – Create a structured outlineThen, it will launch a prompt like: “Create an outline with H2 and H3 headings for an SEO article about social media strategies for SMEs in Spain.” You’ll get an index with sections such as introduction, key platforms, content types, metrics, and final tips.

Step 3 – Develop the text in partsYou can expand each section one by one for greater control. For example: “Develop the section on content types that generate the most interaction for SMEs in the food sector in Spain.” This way, the model adapts the content to the niche and the geographical context.

Step 4 – Optimize on-page SEOOnce you have the draft, ask Perplexity to review the text with a prompt like: “Analyze this content and suggest natural SEO improvements without resorting to keyword stuffing: ”. The AI ​​will suggest synonyms, keyword relocation, heading refinements, and even readability improvements.

Step 5 – Reinforce with data and sourcesTo increase the article's authority, you can request: "Include updated statistics on social media use by SMEs in Spain, with sources." Perplexity will search for recent data and provide links that you should then verify yourself.

Key features of Perplexity that combine with model choice

In addition to choosing AI, Perplexity offers a good handful of extra features which are worth knowing because they greatly improve the user experience, especially if you work with large projects, teams, or need to maintain some control over the information.

Main modes of useYou have Search mode, ideal for quick queries and direct answers; Research mode, which expands the search and cross-references more sources to give you more complete results; and the Labs area, designed for creative and complex tasks (some functions here may be limited in the free version).

Privacy and data controlIf you activate incognito mode, your conversations aren't saved in your history or on the servers, giving you more peace of mind regarding sensitive topics, although you won't be able to retrieve those chats. At any time, you can see which fonts were used, retry a response to generate a different version, or refine the result with new instructions.

Collaboration and third-party modelsIt's possible to share an entire chat or a specific response via a link, without taking screenshots. Furthermore, you can switch between models like GPT, Gemini, or Claude and, in some cases, take advantage of their "superpowers," such as image editing or generation when the engine allows it.

Customization of fonts and informationThe platform lets you choose whether to draw from the internet in general, academic articles, financial documents, or even forums. With Spaces, you can organize information by project, trip, client, etc., and add conversations to each space to keep everything organized.

Current events and task automationDiscover mode functions as a feed displaying recent news in science, technology, art, sports, and more. It's customizable, so you can filter out categories that don't interest you. You can also schedule tasks for Perplexity to execute a specific prompt at a particular time and day (for example, a daily news summary for your industry).

Files and computer usageYou can upload images, audio, or documents for AI to analyze, summarize, or answer questions about their content. On compatible systems, it can also interact with PC applications, capture what's on your screen, and help you understand it without you having to copy and paste.

Voice and conversationIt features dictation so you can speak instead of typing, respecting pauses and punctuation. The paid version includes a more advanced voice mode with different voice options, languages, subtitles, and speeds, ideal even for practicing languages ​​with real-time corrections.

Pro offer with PayPal and browser useAt times, there have been promotions to get Perplexity Pro free for a year using PayPal, which grants access to many of the advanced features at no cost. Additionally, there's a Chrome extension that integrates AI into the address bar, and the Comet browser itself, which lets you automate tasks like filling out forms, checking emails, summarizing pages, or doing research without having to switch between tabs.

With this whole set of models and functions, Perplexity AI ceases to be just "another AI chat" It becomes a comprehensive assistant for staying informed, working, studying, creating content, and organizing. When you learn to combine Best mode with manual template selection, decide when to use Thinking variants, and leverage Focus, Copilot, Spaces, and scheduled tasks, the feeling of being overwhelmed by so many options disappears, and the tool becomes a very reliable ally for almost any type of digital project.