Skip to main content
  1. Home
  2. Computing
  3. News

Perplexity reveals Computer, and it wants AI agents to do all your work

The new subscription feature works like a conductor, tapping models like Gemini, Grok, and ChatGPT for specific jobs.

Add as a preferred source on Google
Sphere, Photography, Outdoors
Perplexity

Perplexity just launched a feature that lets different AI models collaborate on the same task. Called Perplexity Computer, it taps Gemini, Grok, and ChatGPT 5.2 depending on what you need. But Computer does more than answer questions. It completes tasks. The tool is live today for Perplexity Max subscribers on desktop web and will reach Enterprise Max users soon.

The system runs Opus 4.6 as its core reasoning engine. But for specific jobs, it hands off to specialist models. Gemini handles deep research by creating sub-agents. Grok jumps in for speed on lightweight tasks. ChatGPT 5.2 manages long-context recall and wide searches.

Recommended Videos

Computer also connects to Gmail, Outlook, GitHub, Slack, Notion, and Salesforce. It can write docs, build slides, send emails, and run scheduled tasks while you are offline.

The name Computer nods to history. In the 1700s, human computers divided complex work into pieces. Today, Perplexity Computer does the same thing with software.

You get to pick which models run your subtasks

Perplexity Computer is model agnostic, so the company can swap out engines as better ones appear. But you’re not stuck with the defaults. The system lets you choose which models handle your subtasks.

That control matters as token budgets become a real concern for people using AI at work. If you know one model burns through credits faster than another for a simple job, you can pick the cheaper or faster option.

The approach treats AI less like a single appliance and more like a toolbox. Grab Grok for quick answers, Veo 3.1 if you need video, and Nano Banana for images, all within the same session.

Why running multiple models changes the game

The move challenges the idea that AI models are becoming interchangeable commodities. Perplexity argues the opposite. Models are specializing. Each frontier model genuinely excels at different kinds of work, and a smart system should reflect that.

Think of it like having a team instead of one generalist. Gemini might dig through research better. Opus 4.6 handles the heavy reasoning. ChatGPT 5.2 remembers more context from earlier in the conversation. Let them play to their strengths, and the whole system gets more capable.

Computer chains capabilities together naturally. A single prompt can trigger research, analysis, document creation, email delivery, and scheduled follow-up. You do nothing between steps.

What the shift means for your subscription

If you are a Perplexity Max subscriber, you can try the Computer feature today. Enterprise Max users will get access soon. The launch gives you a reason to revisit your subscription and test whether orchestrating multiple models actually saves time or money.

Keep an eye on how the model roster changes. Perplexity built this to be flexible, so the lineup will likely evolve as new models drop. The real test is whether you notice the difference.

If the system picks a faster model for simple searches and a deeper one for research, your workflow should feel smoother without you having to think about it.

Paulo Vargas
Paulo Vargas is an English major turned reporter turned technical writer, with a career that has always circled back to…
This app gives your Mac a music player you’ll actually enjoy using
Apple Music on the Mac is a chore. Liqoria is the fix, and it plays nice with Spotify and YouTube too.
Liqoria music player

The Apple Music app on the Mac is not up to the mark. I don’t like how it looks or behaves, and Apple should take some inspiration from Spotify to make the app more modern and useful. Until that happens, we are stuck with a subpar app experience.

That’s why I never use the Apple Music app on my Mac and rely on third-party apps that let me control music. Today I am featuring one of the latest apps I discovered. It’s called Liqoria, and it’s probably the only music player app you will ever need.

Read more
Anthropic is giving away Claude Fable 5 at no extra cost for a limited time
You can try Claude Fable 5 for free - but don't wait too long
Electronics, Mobile Phone, Phone

Anthropic is making it a little easier for paying subscribers to try its newest AI model without spending extra money. The company has announced a limited-time promotion that gives users on eligible paid plans access to Claude Fable 5 at no additional cost until July 19, 2026. There's a catch, though: the model isn't completely unlimited, and once you hit a usage threshold, you'll either need to start paying or switch to another Claude model.

The promotion comes as competition in the AI assistant market continues to intensify. OpenAI, Google, Microsoft and Anthropic are all racing to get users onto their latest flagship models, often using free trials or promotional access to encourage adoption. Rather than offering unlimited access, Anthropic is betting that giving subscribers enough time to experience Fable 5's capabilities will convince many to continue using it after the promotion ends.

Read more
Scammers are now cloning trusted news websites to steal your money
Breaking news: That breaking news probably isn't breaking news
Scammers are turning trusted news brands into investment traps

Seeing a story on the website of a trusted news organisation is usually enough to lower your guard. Cybercriminals know that, and they're increasingly exploiting the credibility of major publishers to steal money from unsuspecting readers. The latest example involves fake Guardian articles featuring billionaire Jim Ratcliffe. Still, the scam is part of a much larger campaign that's also impersonating the BBC and other well-known media outlets.

According to The Guardian, fraudsters are creating convincing clones of legitimate news websites and filling them with fabricated stories designed to lure readers into bogus cryptocurrency and investment schemes. Instead of trying to hack victims directly, the scammers first convince them they're reading real journalism.

Read more