Perplexity took the world by storm and became everyone's research assistant. I think it has the potential to become a complete AI assistant. Here's why:
What We Actually Use AI Assistants For?
Let's start with what we typically need from AI assistants:
- Access to really good AI models for brainstorming tasks
- Ability to research topics on the internet
- Ability to do some data analysis - Ability to create images
ChatGPT provides all of this and is still a very good default choice.
But lately, the new changes to the product has caused some friction in my workflow. I have found a degradation in the user experience and the model output.
Partially, I feel that it's because of the reduced set of models (since the unification with GPT-5) and the addition of the automatic model switcher (that switches between GPT-5.1 and GPT-5.1-thinking). These changes have made conversing with ChatGPT slow and clunky.
So it's seriously worth considering Perplexity.
What's great about Perplexity?
1. It provides all the goodies that ChatGPT comes with
Perplexity isn't just a search tool anymore. It's evolved into a full AI platform that includes:
Research and Search - This is what Perplexity became famous for, and it's still pretty good at: - Real-time web search on every query - Citation-first approach with transparent sources - Research-oriented app design
Code Interpreter through Labs - Perplexity recently launched a code interpreter through their "Labs" offering.
I have to be honest here, at the moment this feature is a little clunky, whereas ChatGPT's code interpreter is by far the best and a great way to quickly analyze data. This is one area where ChatGPT still has the edge.
Image Generation - Available in the app, covering the creative use cases you'd typically use ChatGPT for. Additionally, it provides several model providers, so you can chose either chatgpt's autoregressive image generator or gemini's diffusion based nano-banana.
2. Access to several model providers
This is what I like most about Perplexity (although others may find this a little overwhelming). You're not locked with a single model. You get access to multiple top-tier models such as:
Claude Sonnet 4.5 - In my personal experience, this is the most value-for-money text-based model out there at the moment. Excellent for coding and technical reasoning, with an optional reasoning toggle for complex analytical tasks.
Google's Gemini 3 Pro - Showing impressive performance lately. It's Google's latest with state-of-the-art multimodal capabilities (text, images, audio, video) and reasoning always enabled.
Other options - If you're a fan of the alternative model providers, those are available too:
- Kimi K2 Thinking (Open-source reasoning model with step-by-step explanations)
- GPT-5 ( the model that powers ChatGPT)
- Grok 4 (X.AI's latest models which are less constrained )
It's important to keep in mind that Perplexity is a model aggregator. When you ask a question, you select which AI processes it.
Need coding help? Use Claude Sonnet 4.5.
Need multimodal analysis? Switch to Gemini 3 Pro or GPT-5.1.
This flexibility alone makes it compelling for me because several models excel at different tasks and it's nice to have the ability to switch between them. I also am biased here because I just like having choices in general.
3. Perplexity's AI browser - Comet
Perplexity launched their AI browser several months before ChatGPT's acclaimed Atlas. In fact, all the things chatgpt boasts about with Atlas are available within comet. On this instance, I felt that Atlas was a reactive measure by OpenAI, seeing the momentum of Perplexity, but we'll never know.
Anyway, their Comet browser is a really handy way of using Perplexity. In fact, I don't use Perplexity through their website. I use it exclusively through Comet and use Firefox for standard browsing. This way I can keep using the browser like I always did, and jump into comet for more boosted activites like AI research.
Comet Features
- Conversational Browsing - Ask questions about what you're reading, compare information across sources, or get summaries all without leaving the page
- Session Memory - The AI maintains context across all your tabs
- Tab Control - When active, Comet can directly interact with the tab (you'll see a blue outline)
- Privacy-first - Sensitive context is processed locally on your machine
The above features are available with a Pro subscription which costs the usual $20 a month. But, recently perplexity made some tie-ups which could make you eligible for a free pro account.
You could be eligible for a free subscription through:
- Student/Teacher offer - If you're a student or educator, you can avail the free account with your university email. Check if you qualify
- Limited time packages - For example, with an Airtel phone or internet subscription (subject to location and subscription tier)
If you're eligible for a free subscription, you should absolutely take advantage of it.
Honest Downsides
Perplexity is not perfect. Here's what it doesn't do as well:
Code Interpreter is clunky - ChatGPT's code interpreter is far superior for data analysis. If you want frictionless analysis of your files or want to do serious data work, ChatGPT has the edge here with a much better and faster code interpreter.
Coding tasks - While Perplexity can write code (and it's good if you select Claude Sonnet 4.5), for serious coding projects with complex debugging and iterative development, you're better off using Claude Code directly.
Frequently Asked Questions
How good is Perplexity's code creation compared to ChatGPT?
Perplexity is a model aggregator, so code quality depends on which model you select. If you choose Claude Sonnet 4.5, you get excellent code, the same quality you'd get using Claude directly. However, if coding is your primary goal, you should use Claude Code directly. It's optimized for serious development work with better debugging and iteration support.
Q: What about the code interpreter for data analysis?
A: ChatGPT's code interpreter is currently better. It's more polished and reliable for quick data analysis. If data analysis is a major part of your workflow, this is one area where ChatGPT still has the edge.
Is Comet actually useful?
I personally still like using the standard browser. But while I use Firefox for normal browsing, I've started using Comet when I need to use perplexity (instead of using it in firefox because it's faster than opening perplexity.ai in a tab). The AI integration is native and a good example of agentic AI. The voice assistant is very cool but a little buggy at the moment.
Which models should I use for what tasks?
My personal recommendation is Claude Sonnet 4.5 for text-based tasks, coding, and technical reasoning (make sure to toggle the 'enable reasoning' switch). Gemini 3 Pro has shown potential for multimodal tasks (when you need to analyze images, audio, or video). GPT 5.1 is a good default choice. The beauty of Perplexity is you can switch between them based on your needs.
The Bottom Line
I'd now go as far as to suggest that Perplexity could be a complete ChatGPT replacement, especially if you can get hold of the pro plan in which case it is hands-down better than your free ChatGPT account.
The Pro version covers most use cases and in some ways (model selection, search quality), it does it better than the heavy-weights.
Perplexity is an AI Pilots recommended product, and I'd suggest you give it a shot. Especially if you are eligible for a free subscription, in which case you should absolutely take advantage of it.
