The $200 AI Question: Should You Upgrade to ChatGPT Pro?

Two hundred dollars is a lot for a single AI tool, especially when alternatives like Anthropic's Claude are available for much less. What could possibly justify a 10x cost?

We spent weeks with o1 pro — the most advanced AI model that comes with ChatGPT Pro — to answer this question. We fed it several books' worth of data, used it for everything from analyzing sales call transcripts to technical SEO audits, and compared it to its sibling o1 and archrival Claude. 

Believe it or not, we see a world where o1 pro's cost is worth it. Of course, it depends on what you plan to use it for — and if you're willing to put up with its annoying limitations.

Match Your Marketing Work With the Right AI Personality

Before we look at whether o1 pro is worth its premium price tag, you need to understand something about AI models: they have distinct personalities.

Wharton professor and innovation expert Ethan Mollick observes that GPT-4 behaves like a focused, analytical workhorse that "wants to get stuff done for you" without much small talk. In our tests, OpenAI's o1 models show these same traits.

Anthropic's Claude 3.5 Sonnet model is warmer and more creative — it's better at grasping nuance, mimicking styles, and evaluating designs.

These personality differences affect how useful each model will be for your specific content marketing tasks.

What to Use o1 pro for

O1 pro has unique capabilities that benefit specific workflows. Rather than treating it as a general-purpose upgrade for all your work, it's important to match these capabilities to your needs.

If your work is analytical, strategic, and involves large volumes of data, o1 pro's capabilities could justify its premium price. But if you mainly focus on writing and other creative tasks, you're better off paying for Claude.

Processing Large Documents and Datasets

O1 pro — just like the normal o1 model * — has an expanded context window (200K tokens, roughly equivalent to a 150-page book) and a large output window (100K tokens, compared to 4,096 tokens for GPT-4 and 8,192 for Claude's 3.5 Sonnet).

These characteristics make o1 pro really useful for processing large volumes of information that other models struggle with. We found it can:

• Identify speakers in a 60-minute podcast transcript and reformat the entire text accordingly.
• Go through hours of meeting notes to extract and categorize all actions for all attendees.
• Reformat or extract information from large CSV files.
• Copy edit long-form content (including newsletters, articles, and eBooks).

None of these tasks are particularly complex, but other models struggle as the volume of information grows — mainly because their limited output windows prevent them from delivering complete results, even when they understand what needs to be done.

_{* While the context window sizes are the same for these two models, o1's responses seem less comprehensive, accurate, and detailed than o1 pro's, especially when the volume of input increases. This likely has to do with o1 pro's ability to do more in-depth analysis (it can "think" longer) and its superior reasoning capabilities.}

Advising and Guiding

O1 pro shines when it acts as an editor, advisor, or coach. We fed it briefs, strategies, and drafts; it found flaws, insights, and improvements worthy of an expert review.

Here are some ways in which we used these capabilities:

• Get structural feedback on drafts.
• Find angles and outlines for content (based on an interview or voice memo transcript).
• Coach podcast show hosts on better interview skills.
• Help us get unstuck when creating scripts and apps (like our content refresh tool Revive).

In most of these cases, we actually used o1 pro's advice as input for Claude.

For example, we first asked o1 to outline a LinkedIn post, based on its analytical strengths. We then gave that outline to Claude for writing a first draft, taking advantage of its creative capabilities. This approach plays to each model's personality strengths.

Analyzing Complex Information

O1 pro has the most advanced reasoning capabilities of current publicly available models. It scores highest on a range of advanced benchmarks, including human-level tests in math (AIME 2024) and coding (Codeforces), as well as AI-specific evaluations like the GPQA Diamond test for PhD-level science questions.

Combine these advanced reasoning capabilities with o1 pro's ability to handle large volumes of information, and you get an AI assistant that really starts to feel more like a peer than an intern.

Sales discovery calls

We always create tailored proposals for potential customers after a first sales discovery call. We already use Perplexity and Claude for this process, and those tools save us 10-20% of time, mostly in speeding up research and drafting.

The o1 pro model blew our minds when we tested it on a sales call transcript: it essentially put together a draft strategy for the customer and made a very nuanced analysis of the client's requirements, including "unspoken" needs it picked up on in the call transcript.

Content audits

Sara Coggin, Associate Head of Content here at Animalz, tested o1 pro for different components of technical SEO audits, like internal link analysis and creating a link graph visualization.

She found it could guide her through sophisticated processes that would normally require specialized SEO expertise: "I'm now confident I can do technical tasks like analyzing the link structure of a website in Python in less than one hour, compared to not being able to do this at all before."

In her testing, the difference between o1 pro and the standard o1 in ChatGPT's $20 plan wasn't dramatic. While o1 pro's recommendations were somewhat more detailed — like suggesting specific Python libraries for creating visualizations — the basic analytical capabilities were similar.

Interview preparation

We're about to relaunch the Animalz podcast (stay tuned!), and o1 pro helped us prepare for interviews.

Following our principle of always starting with human input (as explained in Stay Strong: Never Let AI Fill Your Blank Page), we first prepared our own show goals and topic ideas. We then fed o1 pro extensive background materials on each guest, including complete articles and transcripts from previous podcast appearances.

This comprehensive context allowed the model to suggest more nuanced questions, like:

• What markers do you look for to determine if a marketing task or process is ready for AI automation versus requiring more human touch?
• You've been vocal about AI's impact on marketing efficiency. Where do you see the role of marketing leaders evolving as AI takes on more tasks?
• You've built a massive LinkedIn following—can you walk us through your typical workflow from ideation to posting, and how AI factors into that process?
• When adopting AI tools, which part of the marketing or sales process is typically the trickiest to overhaul, and why do you think that is?

These questions feel a bit stiff — fitting o1's analytical personality — but they also showcase its strengths: focusing on frameworks and decision-making processes, prompting specific examples and workflows, and spotting patterns across our guests' experiences.

After we rephrased these questions in our own voice, they sparked lively, in-depth conversations and lots of insights.

Drawbacks You Should Know About

You might expect that paying $200 per month would guarantee a smooth experience, but that's not the case with o1 pro. Even if you were nodding along and getting excited about the capabilities and use cases we just went through, the following constraints might give you second thoughts.

You can't attach text files OF ANY KIND

You can't upload text files, PDFs, or CSVs directly (images are supported). This limitation means extra steps for every document you want to include: open it in a text editor, select all, copy, and then paste into o1 pro.

With multiple documents, you'll repeat this for each file — and then face the challenge of organizing thousands of words in a cramped chat interface.

It can be long-winded, very long-winded

o1 pro likes to be verbose. It will give a 1,000 word response where three paragraphs would do. During our testing, we frequently had to reply with "be more concise" or "give me the key points only" to get a short response.

It likes to play advisor

o1 pro prefers giving advice instead of telling you directly what to do. For example, when we asked it to review an article outline, it came back with advice like "consider mentioning SEO or brand marketing here."

Sometimes that's what you want, but often you just want a clear answer. Explicitly telling it to give an instruction, decision, or recommendation usually does the trick.

It's slooooooowww

Perhaps most surprisingly, o1 pro can be very slow. Certain requests took several minutes to complete; great thinking capabilities come with great waiting times.

Of course, the longer processing times lead to better results, so it's a bit unfair to complain about them. Still, be prepared to wait when you give it a complex assignment.

It's not private

o1 pro's data privacy features are limited (except for Enterprise plan users). OpenAI can use your inputs for model training unless you specifically opt out or delete chats.

How worried should you be about this? We haven't found any credible, widely reported incident in which — through normal usage* — private data typed by one user became publicly accessible to other users of ChatGPT.

That said, the absence of documented or confirmed cases doesn't mean the risk is zero. If you're concerned about data privacy, you should still treat ChatGPT with caution.

_{* There has been a big data breach in March 2023 that exposed private ChatGPT data for several hours. However, this is different from the idea that you putting in private data would lead to someone else seeing that information in their answers.}

Its videos are no good

O1 pro also gets you access to Sora, OpenAI's video generation model, and Sora sucks isn't great.

While it's probably not the sole reason you'd pay $200 p/m, our brief tests to create any kind of decent video were disappointing.

Should You Pay $200 per Month for ChatGPT Pro?

After weeks of testing, here's a straightforward checklist for you to decide if o1 pro is worth its $200 monthly price tag:

1. You have complex analytical needs.
2. You regularly deal with large volumes of information.
3. You can live with its limitations.
4. You want a specialized tool to complement Claude's creativity, not replace it.

At $2,400 per year, o1 pro isn't a casual purchase, and it's not for everyone.

Testing with simple analytical tasks showed only modest improvements over the $20 version — o1 pro's value comes from handling large amounts of information or tackling multi-layered analysis.

If that doesn't describe your daily work, save your money and stick with the standard tier or alternative models like Claude.

Tim Metz

Follow Tim

Tim is the Director of Marketing and Innovation at Animalz. He's also the creator of Lifeline, a pomodoro timer app for macOS. Tim lives in Koh Samui, Thailand. 🏝️