Google Gemini Advanced vs ChatGPT Plus – Which One Is Better?

I’ve been waiting to get my hands on Google’s latest and greatest LLM, Gemini Ultra…this is now available as part of Google Gemini Advanced (the paid version and a successor to Google Bard). I decided to upgrade my existing Google One subscription and put Ultra / Advanced through its paces…here’s a quick video speedrun:

So, How Does Google’s Best Paid AI Model Perform?

The first thing you notice about Google Gemini is that it is slick and blazingly fast, which is to be expected with the amount of raw compute power Google is able to throw at this. Here’s what I loved and hated about Google Gemini Advanced, after my initial testing (still early days, so these may opinions will likely change soon):

Google Gemini Ultra 1.0 (Advanced) Pros:

• Speed: As I said above, the output is very quick and produced at one go, compared to ChatGPT’s line by line output.
• Integration With Google Workspace: I think this is the best thing working in favour of Google Gemini – the ability to tap into your Gmail (emails) and Google Drive documents is a massive help and probably the most useful feature for me. Natural language search is a game changer, as you can see from the video demo above.
• Access to Google’s Search Ecosystem: Another huge advantage is being able to tap into the quality output from Google Search, News, Flights, Hotels and much much more, which can’t be said for ChatGPT (despite the tie up with Microsoft and access to Bing, though some plugins do a decent job).
• Language and Translation Proficiency: Leveraging Google’s extensive work in language models and translation services, Google Gemini can provide more accurate and nuanced language processing.
• Advanced Capabilities: Google boasts that Gemini Ultra (Advanced) is better at reasoning and designed for complex tasks, including the ability to code better. I haven’t given this a proper test yet, but early 3rd party reviews seem to confirm this.

Google Gemini Cons (and Fails):

• Restricted File Handling: Google Advanced is unable to directly process CSV files or PDFs, hindering its application in data analysis and document processing. This is a huge drawback compared to ChatGPT.
• Subscription Cost: After a two-month free trial, the Google One AI Premium model costs roughly $21 per month (not available as an annual plan), which isn’t justified given its current limitations. I think this could change when Google rolls out Gemini 1.5 with multi-modal capabilities. Perhaps this is the reason Google has also chosen to give everyone a free 2 month trial.
• Lack of Plugins and Custom GPTs: The one thing I love about ChatGPT is the ability to create custom GPTs and use GPTs created by others – it really levels up the tool’s capabilities and is a great first step towards having automated AI agents.
• Image Processing: It’s supposed to be capable of interpreting images, though in my early tests, accuracy was pretty poor. Plus Google has temporarily halted the model’s ability to produce images of people, after it ran into controversies for overcompensating for racial diversity (see news screenshot below)

How Does ChatGPT Plus Compare To Google Gemini Advanced?

Given it has been out longer and has had more time to mature, ChatGPT Plus is a much more versatile and reliable tool for a broader range of tasks, including programming, file processing, and comprehensive text-based interactions.

Key ChatGPT Plus Strengths:

• Advanced Reasoning and Creativity: ChatGPT 4 excels in generating creative content, editing, and iterating with users on various writing tasks. It demonstrates superior problem-solving abilities (especially when using Advanced Data Analysis and domain specific expert plugins like Wolfram Alpha) and broader general knowledge​​.
• Visual Input Capability: A notable feature is its ability to accept and interpret images, generating captions, classifications, and detailed analyses, thus supporting creative and technical tasks involving visual components​​​​.
• Custom GPTs and Plugins: As I said above, the ability to create your own GPTs and access those from others from a rapidly growing library is a game changer. Here’s a great example showing how you can create Powerpoint and Video files easily using these custom GPTs.
• Other: ChatGPT Plus now supports a range of languages, enhancing its accessibility and usability across different linguistic groups​​. Significant improvements have also been made in making ChatGPT 4 safer and more aligned with user expectations, reducing the likelihood of generating disallowed content and increasing factual accuracy​​.

ChatGPT Plus Weaknesses:

• Complexity for Casual Users: The vast array of features and the sophisticated nature of its capabilities might be overwhelming for casual or non-technical users.
• Web Search: While ChatGPT can leverage Bing search to access web content, the overall experience is pretty slow and underwhelming.
• Usage Limitations: I’d love the ability to use longer form content (Anthropic’s Claude 2.0 accepts much more input for example, and Gemini 1.5 Pro promises token lengths in the millions – see note below). Plus the 40 messages every 3hours is a limit that can get pretty annoying when you’re in the flow and trying to get work done.
• Cost: Access to the full range of features, including advanced versions and visual inputs, requires a paid subscription that costs $20 per month, potentially limiting its accessibility to a wider audience.

What’s Next: Google Gemini 1.5 Pro Will Push The Limits

Google DeepMind has also released early demos of what its newest model (built multimodal from the ground up) will be able to do, and some of the capabilities are absolutely amazing.

Gemini 1.5 is described as one of the most advanced AI language models to date, capable of understanding and remembering a vast amount of information—like reading 7.5 million words or about 2% of Wikipedia—and accurately recalling facts.

However, processing such large amounts of data can slow it down a bit. For now, only a few developers and businesses can use it, but improvements are expected to make it faster.

Through demonstrations, like pulling precise details from a long document (demo below) or video (demo above), Gemini 1.5 has shown it’s not just good with words but also with pictures and sounds, outperforming its earlier version.

With the upcoming Gemini 1.5 Ultra, we can expect even more advancements, making it a very promising tool in the future.

Conclusion

It’s been fun testing Google’s latest and greatest AI product – Gemini Advanced 1.0 (Ultra) is certainly powerful, and even more so due to its integration with the Google ecosystem. Gemini 1.5 Pro & Ultra will change the game when they are released, especially with multimodal capabilities and the ability to take in much longer input.

ChatGPT Plus remains king of the pack for now, given all the reasons outlined above. Though I think we’ll be seeing both OpenAI and Google DeepMind battle it out quite aggressively in the months and years ahead…and we’ll be so much better for it!